Datasets in the Art of Stat app

Variables

County

One of the 67 counties in Florida.

Perot VotesQuantitative

Number of votes cast in the county for Ross Perot, the Reform Party candidate in 1996.

Buchanan VotesQuantitative

Number of votes cast in the county for Pat Buchanan, the Reform Party candidate in 2000.

Variables

LakeCategorical: Hancock, Oklawaha, Trafford, George

The name of the lake the gator was caught in.

GenderCategorical: Male, Female

The gender of the alligator.

SizeCategorical: Small, Large

The length of the alligator, here recorded as "Small" if the gator was less than 2.3 meters, and "Large" if the alligator measured more than 2.3 meters.

FoodCategorical: Fish, Invertebrate, Reptile, Bird, Other

The primary food choice of the alligator.

Try it in the Explore Data module

• One Categorical Variable: Select Food as the variable and obtain a frequency table and a bar chart to study the distribution of primary food choice of alligators. Which primary food type was the most common? You can also obtain a pie or donut chart on the third tab.
• Compare Groups: Categorical Response: Find the distribution (conditional proportions) of primary food choice in each of the four lakes. Select Lake as the grouping variable and Food as the response variable. Obtain and compare the conditional proportions provided on the second tab, and visualize the conditional distributions in a side-by-side or stacked bar chart on the third tab.
• Two Categorical Variables: Is there evidence of an association between gender and primary food choice? Select Gender as the row variable and Food as the column variable. Obtain the contingency table and the conditional proportions on the second tab, and look at a side-by-side or stacked bar chart on the third tab. Repeat the analysis, using Size instead of Gender.

Variables

Animal

Identification number of the animal, from 1 to 24.

FamilyCategorical: Canidae, Felidae, Mustelidae

The family name of the animal (Canidae, Felidae, Mustelidae).

Common Name

The name under which the animal is commonly known.

Gestation (days)Quantitative

Average duration of gestational period, in days.

Longevity (yrs)Quantitative

Maximal life span, in years.

Birth weight (g)Quantitative

Average birth weight, in grams.

Adult weight (kg)Quantitative

Average adult weight, in kilograms.

Growth rate (per day)Quantitative

Postnatal growth-rate coefficient, in units of 1/day, from a Gompertz curve. A coefficient of 0.01 means it takes about 310 days to grow from 10% to 90% of adult weight, and that, at its fastest, the animal gains about 0.37% of its adult weight each day. That percentage is the coefficient divided by e. In this dataset the lowest value is the cougar, 0.0061: about 510 days to grow from 10% to 90% of adult weight, and about 0.22% of adult weight per day at its fastest (about 140 g a day for a 63 kg adult). The highest value is the least weasel, 0.0546: about 56 days, and about 2% of adult weight per day at its fastest (about 10 g a day for a 0.5 kg adult).

Try it in the Explore Data module

• Two Quantitative Variables: Try to analyze the relationship between birth weight (x-variable) and adult weight (y-variable), or between gestational period (x-variable) and maximum longevity (y-variable). Select the animals' common name as an ID variable, so you can tap on a dot in the scatterplot and learn which animal it corresponds to. Select Family as the variable to color the dots, so you can see how families differ.

Variables

Student

Sequential numbers for the students, serves as an ID.

FroshCategorical: Yes, No

Whether or not the student was a freshman.

ChocolateCategorical: dark, milk, white

Preferred type of chocolate (dark, milk, white).

Textbook Costs ($)Quantitative, in USD

Amount of money student spent on textbooks in one semester.

Streaming (min/day)Quantitative, in minutes per day

Average amount of time student spent streaming content, per day.

Social Media (min/day)Quantitative, in minutes per day

Average amount of time student spent on social media, per day.

GenderCategorical: Male, Female

Gender of student.

Political ViewCategorical: Conservative, Moderate, Liberal

Political view of student, classified as conservative, moderate or liberal.

Try it in the Explore Data module

• One Categorical Variable: Get a frequency table and a bar chart to study the distribution of preferred type of chocolate. For this, select Chocolate as the variable. Which type of chocolate was the most preferred? You can also obtain a pie or donut chart on the third tab.
• Compare Groups: Categorical Response: Find the distribution (conditional proportions) of political view for freshmen and compare it to the one for upper-classmen. Select Frosh as the grouping variable and Political View as the response variable. Obtain and compare the conditional proportions provided on the second tab, and visualize the conditional distributions in a side-by-side or stacked bar chart on the third tab.
• Two Categorical Variables: Is there evidence of an association between gender and political view? Select Gender as the row variable and Political View as the column variable. Obtain the contingency table and the conditional proportions on the second tab, and look at a side-by-side or stacked bar chart on the third tab.
• One Quantitative Variable: What is the distribution of textbook costs? Are there any outliers? Select Textbook Costs as the variable on the first tab, and look at the descriptive statistics on the second tab. A histogram and a boxplot of the distribution of textbook costs appears on the third tab. Use all of these outputs for a complete description of the textbook cost variable.
• Compare Groups: Quantitative Response: Compare textbook costs for freshmen and upper-classmen. Or, contrast the distribution of time on social media for those identifying as conservative, moderate and liberal. For the former, select Frosh as the grouping variable, and Textbook Costs as the response variable on the first tab. Get the descriptive statistics of textbook costs for each of the two groups on the second tab, and side-by-side boxplots, histograms or density plots on the third tab.
• Two Quantitative Variables: Obtain a scatterplot for analyzing the relationship between the amount of time spent on social media with the amount of time streaming content. Select Social Media as the x-variable (explanatory variable) and Streaming as the y-variable (response variable) on the first tab. The descriptive statistics for each variable are shown on the first tab. The second tab shows the scatterplot and gives the correlation coefficient for these two variables. You can also fit a regression line and visualize it on the scatterplot.

Variables

Coral Reef SiteCategorical: Salt Pond, Cocoloba, Tektite

The site of the reef in the Virgin Islands where the larvae were placed. Three shallow coral reef sites were considered: Salt Pond (18.30806698° N, 64.70905197° W), Cocoloba (18.31528002° N, 64.76064996° W) and Tektite (18.309561998° N, 64.72218004° W). The sites were located at depths of 12.5 m (Tektite), 8 m (Cocoloba) and 5.1 m (Salt Pond). These sites were all relatively exposed on the southern shore of St John, with qualitatively similar hydrodynamic environments.

Reef QualityCategorical: Degraded, High Quality

Healthy or degraded.

Enriched SoundscapeCategorical: No, Yes

Whether or not the site received sound from the underwater playback system.

Deploy_DateCategorical: June, July

Month in 2022 when larvae were placed at the reef site.

Distance from SpeakerQuantitative

Distance, in meters, of the cup of larvae from the underwater speaker. Cups were placed at 1, 5, 10, and 30 m. At Salt Pond the speaker played healthy reef sounds, and at Cocoloba a speaker was on but silent. At Tektite, where no speaker was deployed, the same distances were measured from a marked point at the center of the reef.

Larvae AddedQuantitative

Number of larvae added to specific site.

Larvae SettledQuantitative

Number of larvae settled at specific site.

Percent of Larvae SettledQuantitative

Percent of larvae settled (out of the ones added) at specific site.

Variables

YearQuantitative

The year of the temperature and bloom/fruit observation, ranging from 1946 to 2023.

Avg. Spring TemperatureQuantitative

Average temperature in degrees Celsius for the months of March, April and May, for Vienna, Austria.

Days to First BloomQuantitative

The number of calendar days to the first bloom of the elderberry shrub in Austria.

Days to First FruitQuantitative

The number of calendar days to the first fruit of the elderberry shrub in Austria.

Variables

Juror

Sequential numbers for the jurors, serves as an ID.

InstructionCategorical: Standard, New

Whether jurors received the standard or new instruction on how to evaluate eyewitness testimony.

TestimonyCategorical: Weak, Strong

Whether the eyewitness testimony in the trials the jurors were watching was weak or strong.

Instruction/TestimonyCategorical: Standard/Weak, Standard/Strong, New/Weak, New/Strong

Combination of the (standard or new) instructions and the (weak or strong) testimony, giving the four groups of the study in one column.

VerdictCategorical: Guilty, Not Guilty

Guilty or not guilty.

SexCategorical: Female, Male

Sex of the juror.

AgeQuantitative, in years

Age (in years) of the juror.

RaceCategorical: 6 categories

Race of the juror, recorded as White, Black or African American, Asian, American Indian or Alaskan Native, Native Hawaiian or Other Pacific Islander, or Some Other Race.

EducationCategorical: 6 categories

Education of the juror, recorded as Some High School, High School, Some College Credit, Associate, Bachelor, or Postgraduate.

Income BracketCategorical: Under 30K, 30K to 60K, 60K to 90K, Over 90K

Income bracket of the juror, in US dollars.

Try it in the Explore Data module

• One Categorical Variable: Select Verdict as the variable and obtain a frequency table and a bar chart or pie chart to see what proportion of jurors gave a guilty verdict. You will need to follow up this analysis with the one below.
• Compare Groups: Categorical Response: Find the conditional proportion of a guilty verdict for each instruction - testimony combination. Select Instruction/Testimony as the grouping variable and Verdict as the column variable. Obtain and compare the conditional proportions provided on the second tab, and visualize the conditional distributions in a side-by-side or stacked bar chart on the third tab.
• Two Categorical Variables: Is there evidence of an association between the type of eyewitness testimony (weak or strong) and the verdict reached by jurors? Select Testimony as the row variable and Verdict as the column variable. Obtain the contingency table and the conditional proportions on the second tab, and look at a side-by-side or stacked bar chart on the third tab. Repeat the analysis with Instruction as the row variable, where the two conditional distributions differ much more.

Variables

Room TypeCategorical: Entire Home, Private, Shared

One of private room, shared room or entire home.

PriceQuantitative, in USD

The price of the room per night.

Minimum StayQuantitative, in nights

The minimum number of nights required when booking the room.

ReviewsQuantitative, a count

The number of reviews the room has received on Airbnb.

Try it in the Explore Data module

• One Categorical Variable: Select Room Type as the variable and obtain a frequency table and a bar chart or pie chart to see the distribution of room type of Airbnb listings in New York City.
• One Quantitative Variable: Explore the distribution of the price of Airbnb rooms in New York City. Is it skewed? What is the median price? Select Price as the variable on the first tab, and look at the descriptive statistics given on the second tab. Study the various graphs, such as the histogram and the boxplot given on the third tab for a graphical representation of the distribution of Airbnb prices in NYC.
• Compare Groups: Quantitative Response: Does the distribution of the price of Airbnb rooms in New York City differ by the type of room? Select Room Type as the group variable and Price as the response variable on the first tab. Descriptive statistics for each of the three groups are shown on the second tab, and side-by-side boxplots and histograms are shown on the third tab.

Variables

Model

Make and model of the car.

Miles per gallonQuantitative

Fuel efficiency, in miles per gallon.

Weight (in 1000 lbs)Quantitative

Weight of the car, in thousands of pounds, so 2.62 means 2,620 pounds.

CylindersQuantitative

Number of cylinders in the engine (4, 6 or 8).

Displacement (cu.in.)Quantitative

The total volume swept by all the pistons, in cubic inches, which is the usual measure of engine size.

HorsepowerQuantitative

Gross horsepower of the engine.

TransmissionCategorical: Automatic, Manual

Whether the car has an Automatic or a Manual transmission.

Engine TypeCategorical: V-shaped, Straight

How the cylinders are arranged, either V-shaped in two angled banks or Straight in a single line.

GearsQuantitative

Number of forward gears (3, 4 or 5).

CarburetorsQuantitative

Number of carburetor barrels feeding the engine.

Variables

Loan Nr.

Identifier for the loan.

Loan AmountQuantitative, in USD

In USD, between $1,000 and $40,000.

DurationCategorical: 36, 60 months

Whether this is a loan to be paid back over 36 months or 60 months.

Interest RateQuantitative, a percentage

Quoted as a percentage.

GradeCategorical: A, B, C, D, E, F

A grade (from A to F) that lendingclub.com assigned to the loan, reflecting its risk. The grade of A indicates a loan with the smallest risk, while a grade of F indicates a loan with the greatest risk of defaulting.

HomeCategorical: own, mortgage, rent

Whether the person applying for the loan lives in a home without paying any mortgage ("own"), pays a mortgage ("mortgage"), or pays rent ("rent").

Income (in $1000)Quantitative, in 1,000 USD

Annual income (in 1,000 USD) of person applying for the loan.

FICO ScoreQuantitative, 300 to 850

Lower bound on the FICO score for the person taking out the loan. The FICO score is a three-digit number ranging from 300 to 850. It expresses how likely a person is to repay borrowed money based on their credit history.

Try it in the Explore Data module

• One Categorical Variable: Find the distribution of the loan grades, which vary between A and F, by finding a frequency table and a bar chart. For this, select Grade as the variable, and study the frequency table and bar chart or pie chart.
• Compare Groups: Categorical Response: Compare the distribution of loan grades (from A to F) across the three different options of home ownership: Own, Mortgage or Rent. Find the conditional proportions of the grades for each of the three groups, and compare them in a side-by-side bar graph. For this, select Home as the group variable and Grade as the response variable.
• Two Categorical Variables: Are the duration of the loan and its grade associated, or is the grade a loan receives independent of its duration? Using the data in the Online Lending dataset, get the contingency table with Duration as the row variable and Grade as the column variable. Find the conditional proportions in each of the two rows, and visualize them in a side-by-side or stacked bar graph.
• One Quantitative Variable: Explore the distribution of loan amounts. Is it skewed? What is the median loan amount taken out? For this, select Loan Amount as the variable. Descriptive statistics are shown on the second tab, and various graphs, including a histogram with a boxplot, are on the third tab.
• Compare Groups: Quantitative Response: Does the distribution of loan amounts differ among the three groups of home ownership: Own, Mortgage or Rent? To analyze this, select Home as the group variable and Loan Amount as the response variable. Descriptive statistics for each of the three groups are shown on the second tab, and side-by-side boxplots and histograms are shown on the third tab.
• Two Quantitative Variables: Construct a scatterplot to study the relationship between the interest rate of a loan and the FICO score of the applicant. Select FICO Score as the x-variable (explanatory variable) and Interest Rate as the y-variable (response variable) on the first tab. Descriptive statistics for each variable are shown on the first tab. The second tab constructs the scatterplot between these two observations. You can superimpose a regression line or obtain the correlation coefficient. Use the third tab to predict the loan interest rate for a new applicant with a FICO score of 720 points.

Variables

Penguin

Identification number of the penguin, from 1 to 342.

SpeciesCategorical: Adelie, Chinstrap, Gentoo

Penguin species (Adélie, Chinstrap, or Gentoo).

IslandCategorical: Dream, Torgersen, Biscoe

Island name (Dream, Torgersen, or Biscoe) in the Palmer Archipelago (Antarctica).

Bill Length (mm)Quantitative

Length (in mm) of a penguin's bill, measured along the upper ridge.

Bill Depth (mm)Quantitative

Depth (in mm) of a penguin's bill, measured from the top of the bill to the bottom of the bill.

Flipper Length (mm)Quantitative

Length (in mm) of the penguin's flipper.

Body Mass (g)Quantitative

Body mass (in g) of the penguin.

SexCategorical: Female, Male

Penguin sex (Female or Male). Nine penguins have a missing value.

Try it in the Explore Data module

• Two Quantitative Variables: You can study the relationship between any two of the bill or flipper variables, and color the dots according to species, island, or sex, to reveal certain patterns in the data.

Variables

State

The US state.

RegionCategorical: South, West, Northeast, Midwest

The region in the US that the state belongs to.

Student ExpenditureQuantitative

Current expenditure per pupil in fall enrollment for public elementary and secondary schools, for the 2012-13 school year.

SAT ScoreQuantitative

Average 2016 SAT score of all students in the state taking the SAT.

Participation PercentQuantitative

Percent of all students in a state who took the SAT in 2016.

Participation CategoryCategorical: low, medium, high

Participation percent classified as low, medium, or high, using cutoffs of 21% and 50%.

Adjusted Teacher SalaryQuantitative

Average teacher salary, adjusted for the cost of living in the state.

Teacher SalaryQuantitative

Mean annual salary of public elementary and secondary classroom teachers in the state, for the 2015-16 school year.

Student Teacher RatioQuantitative

Fall enrollment divided by the number of public elementary and secondary classroom teachers, in fall 2014.

Verbal SAT ScoreQuantitative

Average score on the verbal component of the 2016 SAT.

Math SAT ScoreQuantitative

Average score on the math component of the 2016 SAT.

Variables

Student

Identification number of the student, from 1 to 171.

GenderCategorical: Female, Male

Female, male

AthleteCategorical: No, Yes

Whether the student is participating in college athletics

Self Esteem AppearanceQuantitative

A composite score on how much self esteem the student derives from their appearance. Obtained through a battery of questions of the form "When I think I look attractive, I feel good about myself." Each question was scored on a likert scale from 1 to 7 (1=strongly disagree, 7=strongly agree), with higher scores indicating higher self esteem, and the score represents the average over the battery of questions.

Self Esteem ApprovalQuantitative

A composite score on how much self esteem the student derives from how others approve of them. Obtained through a battery of questions of the form "I can't respect myself if others don't respect me." Each question was scored on a likert scale from 1 to 7 (1=strongly disagree, 7=strongly agree), with higher scores indicating higher self esteem, and the score represents the average over the battery of questions.

Eating DisorderQuantitative

A composite score measuring signs of mild eating disorders, obtained through a battery of 9 eating behavior questions such as "I throw up after I have eaten." All questions were scored from 1 to 6 (1="never", 2= "rarely", 3= "sometimes", 4="often", 5="usually", 6="always"), and the average score is reported.

Exercise FrequencyQuantitative

Hours per week exercising. Answer to the question "In the past month, on average, how many hours per week did you engage in physical exercise?"

Variables

Participant

The ID of the participant in the study.

SexCategorical: male, female

The sex of the participant

Weight (kg)Quantitative

The weight of the participant in kg.

Type of WalkCategorical: Normal, Teabag, Putey

The type of walk performed by the participant: Normal, or Mr Teabag or Mr Putey imitations.

Energy (kcal/kg/min)Quantitative

Energy expenditure during the walk, measured in kcal per kg body weight per minute (kcal/kg/min).

Variables

Passenger ID

A sequential numbering of passengers.

ClassCategorical: 1st, 2nd, 3rd, Crew

The class the passenger was booked, ranging from first to third class, and with a separate category for the crew.

SexCategorical: Female, Male

Sex of passenger (Male or Female).

AgeCategorical: Adult, Child

Whether the passenger was a child or an adult.

SurvivedCategorical: Yes, No

Did the passenger survive the sinking (Yes or No).

Try it in the Explore Data module

• One Categorical Variable: Find out what proportion of passengers sailed in 1st class to third class, and crew. Select Class as the variable, and look at the frequency table on the second tab, and the bar and pie chart on the third tab.
• Compare Groups: Categorical Response: Find the proportion of passengers that survived by class booked. Were 1st class passengers more likely to survive? What about crew? Select Class as the grouping variable and Survived as the column variable. Find the proportions from the second tab, and study the bar graphs on the third tab.
• Two Categorical Variables: Are sex of the passenger and survival related? Select Sex as the row variable and Survived as the column variable. Obtain the contingency table and the conditional proportions on the second tab, and look at a side-by-side or stacked bar chart on the third tab. Repeat the analysis with Age instead of Sex.

Variables

LakeCategorical: Hancock, Oklawaha, Trafford, George

The name of the lake the gator was caught in.

GenderCategorical: Male, Female

The gender of the alligator.

SizeCategorical: Small, Large

The length of the alligator, here recorded as "Small" if the gator was less than 2.3 meters, and "Large" if the alligator measured more than 2.3 meters.

FoodCategorical: Fish, Invertebrate, Reptile, Bird, Other

The primary food choice of the alligator.

Try it in the Inference module

• Chi-Square Test (GoF): Select Food as the variable to test whether alligators choose the five primary food types with equal frequency. The null proportions on the third tab are all equal (0.2 each), so every expected count is 219 / 5 = 43.8 and the test has df = 5 - 1 = 4. The standardized residuals show which food types are chosen far more or far less often than equal preference would imply. Press "Enter/Modify Null Proportions" to test proportions of your own choosing.
• Chi-Square Test: Select Gender as the row variable and Food as the column variable to test whether primary food choice is independent of gender; this test has df = (2 - 1) * (5 - 1) = 4. Compare the conditional proportions on the second tab, then read the expected counts, the Chi-Squared statistic and the P-value on the third tab, and use the standardized residuals to find the cells driving the result. Choosing Lake or Size as the row variable turns the same analysis into a test of homogeneity: do the four lakes, or small and large alligators, share one distribution of primary food choice? With Lake the table has df = 12 and seven expected counts fall below 5, so check them before trusting the P-value.
• Inference for a Proportion: Select Food as the variable and Fish as the category to study the proportion of alligators whose primary food choice is fish. The second tab gives a confidence interval for that population proportion, and the third tab tests a hypothesized value: p₀ = 0.2 is a natural choice, since that is what equal preference among the five food types would give.
• Inference Two Proportions: Select Lake as the group variable and Food as the response variable, then pick Hancock and Oklawaha as Group 1 and Group 2 and choose Fish under "Compute Proportion For". The second tab gives a confidence interval for the difference in the proportion of fish eaters between the two lakes, and the third tab tests whether that difference is zero. Size works as a group variable too: compare the proportion of small and large alligators whose primary food is Invertebrate, where the difference is large enough to see in the bar chart.

Variables

Animal

Identification number of the animal, from 1 to 24.

FamilyCategorical: Canidae, Felidae, Mustelidae

The family name of the animal (Canidae, Felidae, Mustelidae).

Common Name

The name under which the animal is commonly known.

Gestation (days)Quantitative

Average duration of gestational period, in days.

Longevity (yrs)Quantitative

Maximal life span, in years.

Birth weight (g)Quantitative

Average birth weight, in grams.

Adult weight (kg)Quantitative

Average adult weight, in kilograms.

Growth rate (per day)Quantitative

Postnatal growth-rate coefficient, in units of 1/day, from a Gompertz curve. A coefficient of 0.01 means it takes about 310 days to grow from 10% to 90% of adult weight, and that, at its fastest, the animal gains about 0.37% of its adult weight each day. That percentage is the coefficient divided by e. In this dataset the lowest value is the cougar, 0.0061: about 510 days to grow from 10% to 90% of adult weight, and about 0.22% of adult weight per day at its fastest (about 140 g a day for a 63 kg adult). The highest value is the least weasel, 0.0546: about 56 days, and about 2% of adult weight per day at its fastest (about 10 g a day for a 0.5 kg adult).

Variables

Patient

Identification number of the girl, from 1 to 29.

Weight BeforeQuantitative

The girl's weight before the cognitive behavioral therapy.

Weight AfterQuantitative

The girl's weight after the therapy.

Try it in the Inference module

• Two Dependent Means: Select Weight Before as the measurement 1 variable, and Weight After as the measurement 2 variable (or the other way around), and inspect the descriptive statistics and boxplots/histograms for each variable, and the differences in the weights for each girl between the weight before and the weight after therapy. Let µ₁ and µ₂ be the mean weight before and after the therapy, and let μd = µ₁ - µ₂ be the mean weight difference. You can find a 95% confidence interval for the mean weight difference μd on the second tab and a hypothesis test for this difference on the third tab. These will allow you to determine whether the mean weight difference is significantly different from (or significantly less than) 0, and estimate how large this difference is.

Variables

ID

Identification number of the participant, from 1 to 1,314.

HeavenCategorical: Yes, No

Answer to the question "Do you believe in heaven?" (Yes or No).

HellCategorical: Yes, No

Answer to the question "Do you believe in hell?" (Yes or No).

Try it in the Inference module

• Two Dependent Proportions: Select Heaven as the row variable and Hell as the column variable, and inspect the resulting contingency table and bar charts given on the first tab. Let p₁ be the proportion of people who believe in heaven and p₂ the proportion of people who believe in hell. You can find a 95% confidence interval for the difference p₁ – p₂ in the population proportions on the second tab and a hypothesis test for this difference on the third tab. These will allow you to determine whether the proportion of people who believe in heaven and the proportion of people who believe in hell differ significantly, and estimate how large this difference is.

Variables

Student

Sequential numbers for the students, serves as an ID.

FroshCategorical: Yes, No

Whether or not the student was a freshman.

ChocolateCategorical: dark, milk, white

Preferred type of chocolate (dark, milk, white).

Textbook Costs ($)Quantitative, in USD

Amount of money student spent on textbooks in one semester.

Streaming (min/day)Quantitative, in minutes per day

Average amount of time student spent streaming content, per day.

Social Media (min/day)Quantitative, in minutes per day

Average amount of time student spent on social media, per day.

GenderCategorical: Male, Female

Gender of student.

Political ViewCategorical: Conservative, Moderate, Liberal

Political view of student, classified as conservative, moderate or liberal.

Try it in the Inference module

• Inference for a Proportion: Select Chocolate as the variable and dark as the category to study the proportion of students who prefer dark chocolate, 34 out of 70. The second tab gives a confidence interval for that population proportion, and the third tab tests a hypothesized value: p₀ = 1/3 is a natural choice, since that is what equal preference among the three types would give.
• Inference Two Proportions: Select Gender as the group variable and Chocolate as the response variable, then pick Male and Female as Group 1 and Group 2 and choose dark under "Compute Proportion For". The second tab gives a confidence interval for the difference in the proportion preferring dark chocolate, and the third tab tests whether that difference is zero. Frosh works as a group variable too.
• Chi-Square Test (GoF): Select Chocolate as the variable to test whether students prefer the three types of chocolate equally often. The null proportions on the third tab are all equal (1/3 each), so every expected count is 70 / 3 = 23.3 and the test has df = 3 - 1 = 2. The standardized residuals show which types are preferred far more or far less often than equal preference would imply. Press "Enter/Modify Null Proportions" to test proportions of your own choosing.
• Chi-Square Test: Select Gender as the row variable and Political View as the column variable to test whether political view is independent of gender; this test has df = (2 - 1) * (3 - 1) = 2. Compare the conditional proportions on the second tab, then read the expected counts, the Chi-Squared statistic and the P-value on the third tab. Only 7 of the 70 students call themselves conservative, so two of the six expected counts fall below 5; check them before trusting the P-value, and compare the result with the Permutation Chi-Square Test in the Resampling module.
• Inference for a Mean: Select Textbook Costs as the variable to estimate what a student spends on books in a semester. The second tab gives a confidence interval for the population mean and the third tab tests a hypothesized value. Spending is right-skewed and its standard deviation is about as large as its mean, so this is a good place to point out that the t procedures lean on the sample size of 70 rather than on the data being normal.
• Inference for Two Means: Select Frosh as the Group Variable and Textbook Costs as the Response Variable. Because Frosh has exactly two labels, the app fills in Group 1 and Group 2 for you. The second tab gives a confidence interval for the difference in mean textbook spending between freshmen and upper-classmen, and the third tab tests whether that difference is zero. Streaming and Social Media work as response variables too.
• ANOVA: Comparing Means: Select Political View as the Group Variable and Textbook Costs as the Response Variable to compare mean spending across the three groups. The second tab gives descriptive statistics and side-by-side plots, and the third tab holds the ANOVA table, where F has df = 3 - 1 = 2 and 70 - 3 = 67. The groups are very unequal in size, 47 liberal against 16 moderate and 7 conservative students, so compare the group standard deviations on the second tab before leaning on the equal-variance assumption. Chocolate is a second three-group option.

Variables

Child

Identification number of the child, from 1 to 86.

Before CampCategorical: yes, no

Whether the child knew how to manage their asthmatic episodes at the start of the camp (yes or no).

After CampCategorical: yes, no

Whether the child knew how to manage their asthmatic episodes at the completion of the camp (yes or no).

Try it in the Inference module

• Two Dependent Proportions: Select After Camp as the row variable and Before Camp as the column variable, (you can also do it the other way around), and inspect the resulting contingency table and bar charts given on the first tab. Let p₁ be the proportion of children who answered "Yes", I know how to manage my asthma after the camp has concluded, and let p₂ be the proportion of children who answered "Yes" before the camp started. You can find a 95% confidence interval for the difference p₁ – p₂ in the population proportions on the second tab of the app and a hypothesis test for this difference on the third tab. These will allow you to determine whether the proportion of children answering "Yes" after the camp concluded differs significantly from the proportion of children answering "Yes" before the camp, and estimate how much these proportions differ.

Variables

Juror

Sequential numbers for the jurors, serves as an ID.

InstructionCategorical: Standard, New

Whether jurors received the standard or new instruction on how to evaluate eyewitness testimony.

TestimonyCategorical: Weak, Strong

Whether the eyewitness testimony in the trials the jurors were watching was weak or strong.

Instruction/TestimonyCategorical: Standard/Weak, Standard/Strong, New/Weak, New/Strong

Combination of the (standard or new) instructions and the (weak or strong) testimony, giving the four groups of the study in one column.

VerdictCategorical: Guilty, Not Guilty

Guilty or not guilty.

SexCategorical: Female, Male

Sex of the juror.

AgeQuantitative, in years

Age (in years) of the juror.

RaceCategorical: 6 categories

Race of the juror, recorded as White, Black or African American, Asian, American Indian or Alaskan Native, Native Hawaiian or Other Pacific Islander, or Some Other Race.

EducationCategorical: 6 categories

Education of the juror, recorded as Some High School, High School, Some College Credit, Associate, Bachelor, or Postgraduate.

Income BracketCategorical: Under 30K, 30K to 60K, 60K to 90K, Over 90K

Income bracket of the juror, in US dollars.

Try it in the Inference module

• Chi-Square Test (GoF): Select Instruction/Testimony as the variable to check the random assignment itself. If jurors were spread evenly over the four instruction-testimony combinations, every expected count is 335 / 4 = 83.75 and the test has df = 4 - 1 = 3. The observed group sizes run from 79 to 87, so this is a test you expect to pass. Press "Enter/Modify Null Proportions" to test proportions of your own choosing.
• Chi-Square Test: Select Instruction as the row variable and Verdict as the column variable to test whether the verdict is independent of the instruction the juror received; this test has df = (2 - 1) * (2 - 1) = 1 and no expected count below 30. Compare the conditional proportions on the second tab, then read the expected counts, the Chi-Squared statistic and the P-value on the third tab. Using Instruction/Testimony as the row variable makes it a four-group comparison with df = 3.
• Inference for a Proportion: Select Verdict as the variable and Guilty as the category to study the proportion of jurors returning a guilty verdict, 61 out of 335. The second tab gives a confidence interval for that population proportion, and the third tab tests a hypothesized value.
• Inference Two Proportions: Select Instruction as the group variable and Verdict as the response variable, then pick Standard and New as Group 1 and Group 2 and choose Guilty under "Compute Proportion For". The second tab gives a confidence interval for the difference in the proportion of guilty verdicts between the two instructions, and the third tab tests whether that difference is zero. Testimony works as a group variable too, and there the weak and strong recordings produced nearly the same proportion of guilty verdicts.

Variables

Model

Make and model of the car.

Miles per gallonQuantitative

Fuel efficiency, in miles per gallon.

Weight (in 1000 lbs)Quantitative

Weight of the car, in thousands of pounds, so 2.62 means 2,620 pounds.

CylindersQuantitative

Number of cylinders in the engine (4, 6 or 8).

Displacement (cu.in.)Quantitative

The total volume swept by all the pistons, in cubic inches, which is the usual measure of engine size.

HorsepowerQuantitative

Gross horsepower of the engine.

TransmissionCategorical: Automatic, Manual

Whether the car has an Automatic or a Manual transmission.

Engine TypeCategorical: V-shaped, Straight

How the cylinders are arranged, either V-shaped in two angled banks or Straight in a single line.

GearsQuantitative

Number of forward gears (3, 4 or 5).

CarburetorsQuantitative

Number of carburetor barrels feeding the engine.

Variables

Loan Nr.

Identifier for the loan.

Loan AmountQuantitative, in USD

In USD, between $1,000 and $40,000.

DurationCategorical: 36, 60 months

Whether this is a loan to be paid back over 36 months or 60 months.

Interest RateQuantitative, a percentage

Quoted as a percentage.

GradeCategorical: A, B, C, D, E, F

A grade (from A to F) that lendingclub.com assigned to the loan, reflecting its risk. The grade of A indicates a loan with the smallest risk, while a grade of F indicates a loan with the greatest risk of defaulting.

HomeCategorical: own, mortgage, rent

Whether the person applying for the loan lives in a home without paying any mortgage ("own"), pays a mortgage ("mortgage"), or pays rent ("rent").

Income (in $1000)Quantitative, in 1,000 USD

Annual income (in 1,000 USD) of person applying for the loan.

FICO ScoreQuantitative, 300 to 850

Lower bound on the FICO score for the person taking out the loan. The FICO score is a three-digit number ranging from 300 to 850. It expresses how likely a person is to repay borrowed money based on their credit history.

Try it in the Inference module

• Inference for a Proportion: Select Grade as the variable and A as the category to study the proportion of loans that received the best of the six grades, 62 out of 200. The second tab gives a confidence interval for that population proportion, and the third tab tests a hypothesized value.
• Inference Two Proportions: Select Duration as the group variable and Grade as the response variable, then pick 36 and 60 as Group 1 and Group 2 and choose A under "Compute Proportion For". The second tab gives a confidence interval for the difference in the proportion of grade A loans between the two loan durations, and the third tab tests whether that difference is zero. Home works as a group variable too, comparing any two of mortgage, rent and own.
• Chi-Square Test (GoF): Select Grade as the variable to test whether the six loan grades occur equally often. The null proportions on the third tab are all equal (1/6 each), so every expected count is 200 / 6 = 33.3 and the test has df = 6 - 1 = 5. The standardized residuals show which grades are far more or far less common than equal frequency would imply. Press "Enter/Modify Null Proportions" to test proportions of your own choosing.
• Chi-Square Test: Select Home as the row variable and Duration as the column variable to test whether the length of the loan is independent of the home ownership of the borrower; this test has df = (3 - 1) * (2 - 1) = 2 and no expected count below 6. Compare the conditional proportions on the second tab, then read the expected counts, the Chi-Squared statistic and the P-value on the third tab. Grade as the column variable gives a much larger table, df = 10, where six of the eighteen expected counts fall below 5 because grades E and F are rare.
• Inference for a Mean: Select Interest Rate as the variable to estimate the mean rate charged on these loans. The second tab gives a confidence interval for the population mean and the third tab tests a hypothesized value. Loan Amount and Income are the other quantitative choices.
• Inference for Two Means: Select Duration as the Group Variable and Interest Rate as the Response Variable to ask whether 60-month loans carry higher rates than 36-month loans. Duration has exactly two labels, so the app fills in Group 1 and Group 2 for you. The second tab gives a confidence interval for the difference in mean interest rate and the third tab tests whether that difference is zero.
• ANOVA: Comparing Means: Select Home as the Group Variable and Income as the Response Variable to compare mean borrower income across the three kinds of home ownership. The second tab gives descriptive statistics and side-by-side plots, and the third tab holds the ANOVA table, where F has df = 3 - 1 = 2 and 200 - 3 = 197. The groups are unequal in size, 103 with a mortgage, 74 renting and 23 owning outright, so compare the group standard deviations before leaning on the equal-variance assumption.

Variables

Penguin

Identification number of the penguin, from 1 to 342.

SpeciesCategorical: Adelie, Chinstrap, Gentoo

Penguin species (Adélie, Chinstrap, or Gentoo).

IslandCategorical: Dream, Torgersen, Biscoe

Island name (Dream, Torgersen, or Biscoe) in the Palmer Archipelago (Antarctica).

Bill Length (mm)Quantitative

Length (in mm) of a penguin's bill, measured along the upper ridge.

Bill Depth (mm)Quantitative

Depth (in mm) of a penguin's bill, measured from the top of the bill to the bottom of the bill.

Flipper Length (mm)Quantitative

Length (in mm) of the penguin's flipper.

Body Mass (g)Quantitative

Body mass (in g) of the penguin.

SexCategorical: Female, Male

Penguin sex (Female or Male). Nine penguins have a missing value.

Variables

Student

Identification number of the student, from 1 to 32.

Reaction Time Using PhoneQuantitative

Reaction time, in milliseconds, on the driving task while using a cell phone.

Reaction Time Not Using PhoneQuantitative

Reaction time, in milliseconds, on the driving task without using a cell phone.

Try it in the Inference module

• Two Dependent Means: Select Reaction Time Using Phone as the measurement 1 variable, and Reaction Time Not Using Phone as the measurement 2 variable, and inspect the descriptive statistics and boxplots/histograms for each case, and then for the differences of reaction times that is obtained by subtracting the reaction time when using a cell phone from the reaction time when not using a cell phone. Let µ₁ and µ₂ be the mean reaction times in the population, and let μd = µ₁ - µ₂ be the mean difference in reaction times when using a cell phone and when not. You can find a 95% confidence interval for the mean difference μd on the second tab and a hypothesis test for this difference on the third tab. These will allow you to determine whether the mean reaction time when using cell phones differs significantly from the mean reaction time when not using cell phones, and estimate how large this difference is.

Variables

Student

Identification number of the student, from 1 to 171.

GenderCategorical: Female, Male

Female, male

AthleteCategorical: No, Yes

Whether the student is participating in college athletics

Self Esteem AppearanceQuantitative

A composite score on how much self esteem the student derives from their appearance. Obtained through a battery of questions of the form "When I think I look attractive, I feel good about myself." Each question was scored on a likert scale from 1 to 7 (1=strongly disagree, 7=strongly agree), with higher scores indicating higher self esteem, and the score represents the average over the battery of questions.

Self Esteem ApprovalQuantitative

A composite score on how much self esteem the student derives from how others approve of them. Obtained through a battery of questions of the form "I can't respect myself if others don't respect me." Each question was scored on a likert scale from 1 to 7 (1=strongly disagree, 7=strongly agree), with higher scores indicating higher self esteem, and the score represents the average over the battery of questions.

Eating DisorderQuantitative

A composite score measuring signs of mild eating disorders, obtained through a battery of 9 eating behavior questions such as "I throw up after I have eaten." All questions were scored from 1 to 6 (1="never", 2= "rarely", 3= "sometimes", 4="often", 5="usually", 6="always"), and the average score is reported.

Exercise FrequencyQuantitative

Hours per week exercising. Answer to the question "In the past month, on average, how many hours per week did you engage in physical exercise?"

Variables

Student

Identification number of the student, from 1 to 10.

MoviesQuantitative

Number of times the student attended movies during the previous year.

SportsQuantitative

Number of times the student attended sporting events during the previous year.

PartiesQuantitative

Number of times the student attended parties during the previous year.

Try it in the Inference module

• Two Dependent Means: Select Movies as the measurement 1 variable, and Sports as the measurement 2 variable, and inspect the descriptive statistics and boxplots/histograms for each type of event and then for the differences in the number of attendances between these two events. Let µ₁ and µ₂ be the mean attendance for movies and sport events, respectively, and let μd = µ₁ - µ₂ be the mean difference in the number of attendances. You can find a 95% confidence interval for the mean difference μd on the second tab and a hypothesis test for this difference on the third tab. These will allow you to determine whether the mean number of movies attended differs significantly from the mean number of sport events attended, and estimate how large this difference is.

Variables

Word

Identification number of the word in the test bank, from 1 to 2,000.

GMDS SystemCategorical: Correct, Incorrect

Whether the generalized minimal distortion segmentation (GMDS) system identified the word correctly (Correct or Incorrect).

CDHMM SystemCategorical: Correct, Incorrect

Whether the continuous density hidden Markov model (CDHMM) system identified the word correctly (Correct or Incorrect).

Try it in the Inference module

• Two Dependent Proportions: Select GMDS System as the row variable and CDHMM System as the column variable, and inspect the resulting contingency table and bar charts given on the first tab. Let p₁ be the proportion of correct identifications with system GMDS and let p₂ be the proportion of correct identifications with system CDHMM. You can find a 95% confidence interval for the difference p₁ – p₂ in the population proportions on the second tab of the app and a hypothesis test for this difference on the third tab. These will allow you to determine whether the proportion of correct identifications differ significantly between the two systems, and estimate how large this difference is.

Variables

Passenger ID

A sequential numbering of passengers.

ClassCategorical: 1st, 2nd, 3rd, Crew

The class the passenger was booked, ranging from first to third class, and with a separate category for the crew.

SexCategorical: Female, Male

Sex of passenger (Male or Female).

AgeCategorical: Adult, Child

Whether the passenger was a child or an adult.

SurvivedCategorical: Yes, No

Did the passenger survive the sinking (Yes or No).

Try it in the Inference module

• Chi-Square Test (GoF): Select Survived as the variable to test whether those on board were as likely to survive as not. The null proportions on the third tab are equal (0.5 each), so both expected counts are 2201 / 2 = 1100.5 and the test has df = 2 - 1 = 1. With only two categories this is the Chi-Squared form of the one-proportion test, so the statistic here is the square of the z statistic the Proportion app reports. Press "Enter/Modify Null Proportions" to test proportions of your own choosing.
• Chi-Square Test: Select Class as the row variable and Survived as the column variable to test whether surviving the sinking was independent of the class booked; this test has df = (4 - 1) * (2 - 1) = 3. Compare the conditional proportions on the second tab, then read the expected counts, the Chi-Squared statistic and the P-value on the third tab, and use the standardized residuals to find which classes fared better or worse than independence would predict. Sex or Age as the row variable gives the same analysis with df = 1.
• Inference for a Proportion: Select Survived as the variable and Yes as the category to study the proportion of the 2,201 people on board who survived. The second tab gives a confidence interval for that proportion, and the third tab tests a hypothesized value: p₀ = 0.5 is a natural choice.

Variables

Textbook

Identification number of the textbook, from 1 to 10.

Web Site 1Quantitative

Cost of the textbook at Web Site 1, in USD.

Web Site 2Quantitative

Cost of the textbook at Web Site 2, in USD.

Try it in the Inference module

• Two Dependent Means: Select Web Site 1 as the measurement 1 variable, and Web Site 2 as the measurement 2 variable (or the other way around), and inspect the descriptive statistics and boxplots/histograms for each variable, and the differences in cost. Let µ₁ and µ₂ be the mean cost for web site 1 and 2, respectively, and let μd = µ₁ - µ₂ be the mean textbook cost difference. You can find a 95% confidence interval for the mean cost difference μd on the second tab and a hypothesis test for this difference on the third tab. These will allow you to determine whether the mean cost is significantly different between the two web sites, and estimate how large the difference in mean cost is.

Variables

Athlete

Identification number of the athlete, from 1 to 80.

Treatment 1Categorical: cured, not cured

Whether the foot that received Treatment 1 was cured (cured or not cured).

Treatment 2Categorical: cured, not cured

Whether the foot that received Treatment 2 was cured (cured or not cured).

Try it in the Inference module

• Two Dependent Proportions: Select Treatment 1 as the row variable and Treatment 2 as the column variable, (or the other way around), and inspect the resulting contingency table and bar charts given on the first tab. Let p₁ be the proportion of cured feet under Treatment 1, and let p₂ be the proportion of cured feet under Treatment 2. You can find a 95% confidence interval for the difference p₁ – p₂ in the proportions of cured feet on the second tab of the app and a hypothesis test for this difference on the third tab. These will allow you to determine whether the proportion of cured feet is significantly different between the two treatments, and estimate how much these two proportions differ.

Variables

County

One of the 67 counties in Florida.

Perot VotesQuantitative

Number of votes cast in the county for Ross Perot, the Reform Party candidate in 1996.

Buchanan VotesQuantitative

Number of votes cast in the county for Pat Buchanan, the Reform Party candidate in 2000.

Variables

LakeCategorical: Hancock, Oklawaha, Trafford, George

The name of the lake the gator was caught in.

GenderCategorical: Male, Female

The gender of the alligator.

SizeCategorical: Small, Large

The length of the alligator, here recorded as "Small" if the gator was less than 2.3 meters, and "Large" if the alligator measured more than 2.3 meters.

FoodCategorical: Fish, Invertebrate, Reptile, Bird, Other

The primary food choice of the alligator.

Try it in the Resampling module

• Bootstrap Proportion: Select Food as the variable and Fish as the category, then generate at least 2,000 bootstrap samples on the second tab. The third tab turns the bootstrap distribution into a percentile bootstrap confidence interval for the proportion of alligators whose primary food is fish, and lets you compare it to the Wald and Agresti-Coull intervals.
• Permutation Chi-Square Test: Select Gender as the row variable and Food as the column variable, the same setup as the Chi-Square Test in the Inference module, then generate at least 2,000 permutations on the second tab to build the permutation distribution of the Chi-Squared statistic under independence. The third tab locates the observed statistic in that distribution to obtain a P-value. Try Lake as the row variable as well: there seven of the twenty expected counts fall below 5, so the usual Chi-Squared approximation is shaky while the permutation P-value remains valid.

Variables

Animal

Identification number of the animal, from 1 to 24.

FamilyCategorical: Canidae, Felidae, Mustelidae

The family name of the animal (Canidae, Felidae, Mustelidae).

Common Name

The name under which the animal is commonly known.

Gestation (days)Quantitative

Average duration of gestational period, in days.

Longevity (yrs)Quantitative

Maximal life span, in years.

Birth weight (g)Quantitative

Average birth weight, in grams.

Adult weight (kg)Quantitative

Average adult weight, in kilograms.

Growth rate (per day)Quantitative

Postnatal growth-rate coefficient, in units of 1/day, from a Gompertz curve. A coefficient of 0.01 means it takes about 310 days to grow from 10% to 90% of adult weight, and that, at its fastest, the animal gains about 0.37% of its adult weight each day. That percentage is the coefficient divided by e. In this dataset the lowest value is the cougar, 0.0061: about 510 days to grow from 10% to 90% of adult weight, and about 0.22% of adult weight per day at its fastest (about 140 g a day for a 63 kg adult). The highest value is the least weasel, 0.0546: about 56 days, and about 2% of adult weight per day at its fastest (about 10 g a day for a 0.5 kg adult).

Variables

Student

Sequential numbers for the students, serves as an ID.

FroshCategorical: Yes, No

Whether or not the student was a freshman.

ChocolateCategorical: dark, milk, white

Preferred type of chocolate (dark, milk, white).

Textbook Costs ($)Quantitative, in USD

Amount of money student spent on textbooks in one semester.

Streaming (min/day)Quantitative, in minutes per day

Average amount of time student spent streaming content, per day.

Social Media (min/day)Quantitative, in minutes per day

Average amount of time student spent on social media, per day.

GenderCategorical: Male, Female

Gender of student.

Political ViewCategorical: Conservative, Moderate, Liberal

Political view of student, classified as conservative, moderate or liberal.

Try it in the Resampling module

• Bootstrap Mean or Median: Select Textbook Costs as the variable and Population Mean as the Population Parameter of Interest, then generate at least 2,000 bootstrap samples on the second tab. The third tab gives a percentile bootstrap confidence interval for mean textbook spending. Switching the parameter to Population Median is worth doing here, because spending is right-skewed.
• Permutation One Sample: Select Textbook Costs as the variable, then choose Median of Distribution under Hypothesis of Interest and enter a null value; the second and third tabs stay locked until you do. Generate at least 2,000 permutations and read the P-value on the third tab. The mean version of this test assumes the observations come from a symmetric distribution, which textbook spending is not, so the median hypothesis is the sounder choice for this variable.
• Bootstrap Proportion: Select Chocolate as the variable and dark as the category, then generate at least 2,000 bootstrap samples on the second tab. The third tab turns the bootstrap distribution into a percentile bootstrap confidence interval for the proportion of students preferring dark chocolate, and lets you compare it to the Wald and Agresti-Coull intervals.
• Bootstrap Two Samples: Select Frosh as the Group Variable and Textbook Costs as the Response Variable, and keep Difference Between Means as the Population Parameter of Interest. Generate at least 2,000 bootstrap samples on the second tab; the third tab gives a percentile bootstrap confidence interval for the difference in mean spending between freshmen and upper-classmen. Difference Between Medians is the better parameter if the boxplots show how skewed the two groups are.
• Permutation Two Samples: Select Frosh as the Group Variable and Textbook Costs as the Response Variable, choose Equal Group Means under Hypothesis of Interest and Difference in Sample Means as the Test Statistic, then generate at least 2,000 permutations on the second tab. The third tab locates the observed difference in the permutation distribution to obtain a P-value.
• Permutation Chi-Square Test: Select Gender as the row variable and Political View as the column variable, the same setup as the Chi-Square Test in the Inference module, then generate at least 2,000 permutations on the second tab to build the permutation distribution of the Chi-Squared statistic under independence. The third tab locates the observed statistic in that distribution to obtain a P-value. Two of the six expected counts fall below 5 here, so the usual Chi-Squared approximation is shaky while the permutation P-value remains valid.

Variables

Juror

Sequential numbers for the jurors, serves as an ID.

InstructionCategorical: Standard, New

Whether jurors received the standard or new instruction on how to evaluate eyewitness testimony.

TestimonyCategorical: Weak, Strong

Whether the eyewitness testimony in the trials the jurors were watching was weak or strong.

Instruction/TestimonyCategorical: Standard/Weak, Standard/Strong, New/Weak, New/Strong

Combination of the (standard or new) instructions and the (weak or strong) testimony, giving the four groups of the study in one column.

VerdictCategorical: Guilty, Not Guilty

Guilty or not guilty.

SexCategorical: Female, Male

Sex of the juror.

AgeQuantitative, in years

Age (in years) of the juror.

RaceCategorical: 6 categories

Race of the juror, recorded as White, Black or African American, Asian, American Indian or Alaskan Native, Native Hawaiian or Other Pacific Islander, or Some Other Race.

EducationCategorical: 6 categories

Education of the juror, recorded as Some High School, High School, Some College Credit, Associate, Bachelor, or Postgraduate.

Income BracketCategorical: Under 30K, 30K to 60K, 60K to 90K, Over 90K

Income bracket of the juror, in US dollars.

Try it in the Resampling module

• Bootstrap Proportion: Select Verdict as the variable and Guilty as the category, then generate at least 2,000 bootstrap samples on the second tab. The third tab turns the bootstrap distribution into a percentile bootstrap confidence interval for the proportion of guilty verdicts, and lets you compare it to the Wald and Agresti-Coull intervals.
• Permutation Chi-Square Test: Select Instruction as the row variable and Verdict as the column variable, the same setup as the Chi-Square Test in the Inference module, then generate at least 2,000 permutations on the second tab to build the permutation distribution of the Chi-Squared statistic under independence. The third tab locates the observed statistic in that distribution to obtain a P-value. Permuting the instruction labels mirrors the design of the study, in which jurors were randomly assigned to the standard or the new instruction.

Variables

Flight

Flight identifier (1–75).

Delay (min)Quantitative

Delay in minutes.

Variables

Room TypeCategorical: Entire Home, Private, Shared

One of private room, shared room or entire home.

PriceQuantitative, in USD

The price of the room per night.

Minimum StayQuantitative, in nights

The minimum number of nights required when booking the room.

ReviewsQuantitative, a count

The number of reviews the room has received on Airbnb.

Try it in the Resampling module

• Bootstrap Mean or Median: Select Price as the variable and Population Mean as the Population Parameter of Interest, then generate at least 2,000 bootstrap samples on the second tab. The third tab gives a percentile bootstrap confidence interval for the mean nightly price. Because prices are strongly right-skewed, running it again with Population Median shows how differently the two parameters behave on this variable.
• Permutation One Sample: Select Price as the variable, then choose Median of Distribution under Hypothesis of Interest and enter a null value; the second and third tabs stay locked until you do. Generate at least 2,000 permutations and read the P-value on the third tab. The mean version of this test assumes a symmetric distribution, which Airbnb prices are not, so the median hypothesis is the sounder choice here.
• Bootstrap Proportion: Select Room Type as the variable and Private as the category, then generate at least 2,000 bootstrap samples on the second tab. The third tab turns the bootstrap distribution into a percentile bootstrap confidence interval for the proportion of listings that are private rooms, and lets you compare it to the Wald and Agresti-Coull intervals.
• Bootstrap Two Samples: Select Room Type as the Group Variable and Price as the Response Variable. Room Type has three labels, so choose which two to compare under Group 1 and Group 2 — Entire Home against Private is the widest gap. Keep Difference Between Means as the Population Parameter of Interest and generate at least 2,000 bootstrap samples on the second tab; the third tab gives the percentile bootstrap confidence interval for the difference.
• Permutation Two Samples: Select Room Type as the Group Variable and Price as the Response Variable, pick two of the three room types as Group 1 and Group 2, then choose Equal Group Means under Hypothesis of Interest and Difference in Sample Means as the Test Statistic. Generate at least 2,000 permutations on the second tab and read the P-value on the third.

Variables

Model

Make and model of the car.

Miles per gallonQuantitative

Fuel efficiency, in miles per gallon.

Weight (in 1000 lbs)Quantitative

Weight of the car, in thousands of pounds, so 2.62 means 2,620 pounds.

CylindersQuantitative

Number of cylinders in the engine (4, 6 or 8).

Displacement (cu.in.)Quantitative

The total volume swept by all the pistons, in cubic inches, which is the usual measure of engine size.

HorsepowerQuantitative

Gross horsepower of the engine.

TransmissionCategorical: Automatic, Manual

Whether the car has an Automatic or a Manual transmission.

Engine TypeCategorical: V-shaped, Straight

How the cylinders are arranged, either V-shaped in two angled banks or Straight in a single line.

GearsQuantitative

Number of forward gears (3, 4 or 5).

CarburetorsQuantitative

Number of carburetor barrels feeding the engine.

Variables

Loan Nr.

Identifier for the loan.

Loan AmountQuantitative, in USD

In USD, between $1,000 and $40,000.

DurationCategorical: 36, 60 months

Whether this is a loan to be paid back over 36 months or 60 months.

Interest RateQuantitative, a percentage

Quoted as a percentage.

GradeCategorical: A, B, C, D, E, F

A grade (from A to F) that lendingclub.com assigned to the loan, reflecting its risk. The grade of A indicates a loan with the smallest risk, while a grade of F indicates a loan with the greatest risk of defaulting.

HomeCategorical: own, mortgage, rent

Whether the person applying for the loan lives in a home without paying any mortgage ("own"), pays a mortgage ("mortgage"), or pays rent ("rent").

Income (in $1000)Quantitative, in 1,000 USD

Annual income (in 1,000 USD) of person applying for the loan.

FICO ScoreQuantitative, 300 to 850

Lower bound on the FICO score for the person taking out the loan. The FICO score is a three-digit number ranging from 300 to 850. It expresses how likely a person is to repay borrowed money based on their credit history.

Try it in the Resampling module

• Bootstrap Mean or Median: Select Loan Amount as the variable and Population Mean as the Population Parameter of Interest, then generate at least 2,000 bootstrap samples on the second tab. The third tab gives a percentile bootstrap confidence interval for the mean amount borrowed.
• Permutation One Sample: Select Interest Rate as the variable, choose a hypothesis under Hypothesis of Interest and enter a null value; the second and third tabs stay locked until you do. Generate at least 2,000 permutations on the second tab and read the P-value on the third. Mean of Symmetric Distribution assumes the observations are symmetric about the null value, while Median of Distribution allows the skew that interest rates actually show.
• Bootstrap Proportion: Select Grade as the variable and A as the category, then generate at least 2,000 bootstrap samples on the second tab. The third tab turns the bootstrap distribution into a percentile bootstrap confidence interval for the proportion of grade A loans, and lets you compare it to the Wald and Agresti-Coull intervals.
• Bootstrap Two Samples: Select Duration as the Group Variable and Interest Rate as the Response Variable, and keep Difference Between Means as the Population Parameter of Interest. Generate at least 2,000 bootstrap samples on the second tab; the third tab gives a percentile bootstrap confidence interval for the difference in mean rate between 36-month and 60-month loans.
• Permutation Two Samples: Select Duration as the Group Variable and Interest Rate as the Response Variable, choose Equal Group Means under Hypothesis of Interest and Difference in Sample Means as the Test Statistic, then generate at least 2,000 permutations on the second tab. The third tab locates the observed difference in the permutation distribution to obtain a P-value.
• Permutation Chi-Square Test: Select Home as the row variable and Grade as the column variable, the same setup as the Chi-Square Test in the Inference module, then generate at least 2,000 permutations on the second tab to build the permutation distribution of the Chi-Squared statistic under independence. The third tab locates the observed statistic in that distribution to obtain a P-value. Six of the eighteen expected counts in this table fall below 5, so the usual Chi-Squared approximation is shaky while the permutation P-value remains valid.
• Bootstrap Correlation: Select FICO Score as the x-Variable and Interest Rate as the y-Variable, keep Pearson Correlation Coefficient as the Population Parameter of Interest, and generate at least 2,000 bootstrap samples on the second tab. The third tab gives a percentile bootstrap confidence interval for the correlation, whose sample value is r = -0.46. Switching the parameter to Slope of Regression Line gives a bootstrap interval for the slope instead.

Variables

Penguin

Identification number of the penguin, from 1 to 342.

SpeciesCategorical: Adelie, Chinstrap, Gentoo

Penguin species (Adélie, Chinstrap, or Gentoo).

IslandCategorical: Dream, Torgersen, Biscoe

Island name (Dream, Torgersen, or Biscoe) in the Palmer Archipelago (Antarctica).

Bill Length (mm)Quantitative

Length (in mm) of a penguin's bill, measured along the upper ridge.

Bill Depth (mm)Quantitative

Depth (in mm) of a penguin's bill, measured from the top of the bill to the bottom of the bill.

Flipper Length (mm)Quantitative

Length (in mm) of the penguin's flipper.

Body Mass (g)Quantitative

Body mass (in g) of the penguin.

SexCategorical: Female, Male

Penguin sex (Female or Male). Nine penguins have a missing value.

Variables

State

The US state.

RegionCategorical: South, West, Northeast, Midwest

The region in the US that the state belongs to.

Student ExpenditureQuantitative

Current expenditure per pupil in fall enrollment for public elementary and secondary schools, for the 2012-13 school year.

SAT ScoreQuantitative

Average 2016 SAT score of all students in the state taking the SAT.

Participation PercentQuantitative

Percent of all students in a state who took the SAT in 2016.

Participation CategoryCategorical: low, medium, high

Participation percent classified as low, medium, or high, using cutoffs of 21% and 50%.

Adjusted Teacher SalaryQuantitative

Average teacher salary, adjusted for the cost of living in the state.

Teacher SalaryQuantitative

Mean annual salary of public elementary and secondary classroom teachers in the state, for the 2015-16 school year.

Student Teacher RatioQuantitative

Fall enrollment divided by the number of public elementary and secondary classroom teachers, in fall 2014.

Verbal SAT ScoreQuantitative

Average score on the verbal component of the 2016 SAT.

Math SAT ScoreQuantitative

Average score on the math component of the 2016 SAT.

Variables

Student

Identification number of the student, from 1 to 171.

GenderCategorical: Female, Male

Female, male

AthleteCategorical: No, Yes

Whether the student is participating in college athletics

Self Esteem AppearanceQuantitative

A composite score on how much self esteem the student derives from their appearance. Obtained through a battery of questions of the form "When I think I look attractive, I feel good about myself." Each question was scored on a likert scale from 1 to 7 (1=strongly disagree, 7=strongly agree), with higher scores indicating higher self esteem, and the score represents the average over the battery of questions.

Self Esteem ApprovalQuantitative

A composite score on how much self esteem the student derives from how others approve of them. Obtained through a battery of questions of the form "I can't respect myself if others don't respect me." Each question was scored on a likert scale from 1 to 7 (1=strongly disagree, 7=strongly agree), with higher scores indicating higher self esteem, and the score represents the average over the battery of questions.

Eating DisorderQuantitative

A composite score measuring signs of mild eating disorders, obtained through a battery of 9 eating behavior questions such as "I throw up after I have eaten." All questions were scored from 1 to 6 (1="never", 2= "rarely", 3= "sometimes", 4="often", 5="usually", 6="always"), and the average score is reported.

Exercise FrequencyQuantitative

Hours per week exercising. Answer to the question "In the past month, on average, how many hours per week did you engage in physical exercise?"

Variables

Passenger ID

A sequential numbering of passengers.

ClassCategorical: 1st, 2nd, 3rd, Crew

The class the passenger was booked, ranging from first to third class, and with a separate category for the crew.

SexCategorical: Female, Male

Sex of passenger (Male or Female).

AgeCategorical: Adult, Child

Whether the passenger was a child or an adult.

SurvivedCategorical: Yes, No

Did the passenger survive the sinking (Yes or No).

Try it in the Resampling module

• Permutation Chi-Square Test: Select Sex as the row variable and Survived as the column variable, the same setup as the Chi-Square Test in the Inference module, then generate at least 2,000 permutations on the second tab to build the permutation distribution of the Chi-Squared statistic under independence. The third tab locates the observed statistic in that distribution to obtain a P-value. With 2,201 people on board every expected count is in the hundreds, so here the permutation P-value should agree closely with the Chi-Squared P-value.

Variables

County

One of the 67 counties in Florida.

Perot VotesQuantitative

Number of votes cast in the county for Ross Perot, the Reform Party candidate in 1996.

Buchanan VotesQuantitative

Number of votes cast in the county for Pat Buchanan, the Reform Party candidate in 2000.

Try it in the Regression module

• Simple Linear Regression: Analyze the relationship, visualize residuals on the scatterplot, and identify them on the residual plot. (Select County as an ID variable.) Make predictions for the vote total for Pat Buchanan based on the number of votes for Ross Perot in 1996 and a linear regression model.

Variables

Animal

Identification number of the animal, from 1 to 24.

FamilyCategorical: Canidae, Felidae, Mustelidae

The family name of the animal (Canidae, Felidae, Mustelidae).

Common Name

The name under which the animal is commonly known.

Gestation (days)Quantitative

Average duration of gestational period, in days.

Longevity (yrs)Quantitative

Maximal life span, in years.

Birth weight (g)Quantitative

Average birth weight, in grams.

Adult weight (kg)Quantitative

Average adult weight, in kilograms.

Growth rate (per day)Quantitative

Postnatal growth-rate coefficient, in units of 1/day, from a Gompertz curve. A coefficient of 0.01 means it takes about 310 days to grow from 10% to 90% of adult weight, and that, at its fastest, the animal gains about 0.37% of its adult weight each day. That percentage is the coefficient divided by e. In this dataset the lowest value is the cougar, 0.0061: about 510 days to grow from 10% to 90% of adult weight, and about 0.22% of adult weight per day at its fastest (about 140 g a day for a 63 kg adult). The highest value is the least weasel, 0.0546: about 56 days, and about 2% of adult weight per day at its fastest (about 10 g a day for a 0.5 kg adult).

Variables

Patient

ID of the patient included in the study.

Labeling IndexQuantitative

Percentage of labeled cells.

Remission

Whether or not the patient went into remission (1 = Yes, 0 = No).

Try it in the Regression module

• Logistic Regression: Fit the logistic regression model to see (and visualize on the scatterplot) how the probability of remission depends on the labeling index. Predict the probability of cancer remission for a patient with a labeling index of 20.1, the average labeling index of all patients in the study. Estimate by how much the odds of remission increase for every one percentage point increase in the labeling index.

Variables

HoursQuantitative

At which hour was the concentration measured

ConcentrationQuantitative

The free chlorine concentration (in mg per liter)

Try it in the Regression module

• Exponential Regression: Analyze the relationship between the chlorine concentration over time. Estimate by how much the chlorine concentration drops each hour.

Variables

Id

ID of study participant.

AgeQuantitative

Age of participant in years.

SexCategorical: Female, Male

Sex of participant (Female/Male).

Yawn

Whether the participant yawned at least one time when watching the video (Yes or No).

Try it in the Regression module

• Logistic Regression: Are participants less likely to show contagious yawning as their age increased? Fit a logistic regression model using age as a predictor, and find and interpret a confidence interval for its slope. What is the predicted probability of yawning for a participant who is 20 years old? What is the effect on the odds of yawning when age increases by 10 years?

Variables

ID

ID of the person included in the sample.

Annual Income (in Euro 1000)Quantitative

Annual income in Euro 1,000 (so 12 means an income of 12,000 Euros).

Credit Card

Does person have at least one travel credit card (1 = Yes, 0 = No).

Try it in the Regression module

• Logistic Regression: Fit the logistic regression model to see (and visualize on the scatterplot) how the probability of owning a travel credit card depends on income. Predict the probability of owning a card for a person with 40,000 Euros annual income. Find the lower and upper bound of a confidence interval for this prediction. Estimate by how much the odds of owning a credit card change when a person's income increases by 10,000 Euros.

Variables

Customer

Identification number of the customer, from 1 to 10,000.

DefaultCategorical: Yes, No

Whether the customer failed to repay their credit card debt (Yes for 333 customers, No for the other 9,667).

BalanceQuantitative

Average balance left on the credit card after the monthly payment is made, in dollars.

IncomeQuantitative

Income of the customer, in dollars.

StudentCategorical: Yes, No

Whether the customer is a student (Yes or No).

Variables

Day

Day in March, 2020.

Day (since March 5th)Quantitative

Number of days since March 5th, 2020.

COVID CasesQuantitative

Number of confirmed COVID-19 cases in New York City.

Try it in the Regression module

• Exponential Regression: Try to analyze the relationship between the number of days (select "Day (since March 5th)" as the x-variable) and the number of cases (y-variable), to see if the number of cases increased exponentially.

Variables

YearQuantitative

The year of the temperature and bloom/fruit observation, ranging from 1946 to 2023.

Avg. Spring TemperatureQuantitative

Average temperature in degrees Celsius for the months of March, April and May, for Vienna, Austria.

Days to First BloomQuantitative

The number of calendar days to the first bloom of the elderberry shrub in Austria.

Days to First FruitQuantitative

The number of calendar days to the first fruit of the elderberry shrub in Austria.

Variables

Athlete

Identification number of the athlete, from 1 to 64.

AgeQuantitative

Age of the athlete, in years.

Weight (lbs)Quantitative

Total body weight, in pounds.

Height (in)Quantitative

Height, in inches.

Body Fat (%)Quantitative

Percentage of body weight that is fat.

Lean Body MassQuantitative

Body weight excluding fat, in pounds. This is computed from the two previous variables as weight times one minus the body fat proportion.

Repetitions 55Quantitative

Number of bench press repetitions completed before fatigue at a weight of 55 pounds (25 kg).

Repetitions 70Quantitative

Number of bench press repetitions completed before fatigue at a weight of 70 pounds (31.8 kg).

Max Bench PressQuantitative

The heaviest weight, in pounds, that the athlete could bench press a single time.

Variables

Patient

Identification number of the patient, from 1 to 303.

Heart DiseaseCategorical: Yes, No

Whether the patient was found to have heart disease, meaning a major artery narrowed by more than 50% (Yes for 139 patients, No for the other 164).

AgeQuantitative

Age of the patient, in years.

SexCategorical: Male, Female

Male or Female.

Chest Pain TypeCategorical: Typical Angina, Atypical Angina, Non-Anginal Pain, Asymptomatic

The kind of chest pain reported. Typical Angina is the classic pattern brought on by exertion, Atypical Angina fits the pattern only partly, Non-Anginal Pain is chest pain from some other cause, and Asymptomatic means the patient reported no chest pain at all.

Blood PressureQuantitative

Resting blood pressure on admission to the hospital, in mm Hg.

CholesterolQuantitative

Serum cholesterol, in mg/dl.

Blood SugarCategorical: Elevated, Regular

Fasting blood sugar, recorded as Elevated when it exceeds 120 mg/dl and Regular otherwise.

Maximum Heart RateQuantitative

The highest heart rate the patient reached during the exercise test, in beats per minute.

Exercise Induced AnginaCategorical: Yes, No

Whether the exercise itself brought on chest pain (Yes or No).

Old PeakQuantitative

How far the ST segment of the electrocardiogram was depressed by exercise, compared with the patient at rest. Larger values point to less blood reaching the heart muscle.

SlopeCategorical: Upsloping, Flat, Downsloping

The direction the ST segment took at peak exercise, either Upsloping, Flat or Downsloping.

Vessels ColoredQuantitative

How many of the major blood vessels, from 0 to 3, showed up as narrowed when dye was injected and viewed under X-ray.

Thallium ScanCategorical: Normal, Fixed Defect, Reversible Defect

Result of a thallium scan, which traces how well blood reaches the heart muscle. Normal means an even supply, a Fixed Defect is an area that stays poorly supplied even at rest, and a Reversible Defect is an area that runs short of blood during exercise but recovers at rest.

Variables

Crab

ID of the female crab included in the study.

Satellite

Whether the female crab had any male horseshoe crabs (satellites) nesting nearby (yes or no).

Weight (kg)Quantitative

Weight of the female horseshoe crab in kg.

Width (cm)Quantitative

Width of the shell of the female horseshoe crab in cm.

Shell Color

The color of the shell of the female horseshoe crab, acting as a proxy for age. The darker the color, the older the crab.

Shell Color (numerical)

The shell color coded as a number: 1 = medium-light, 2 = medium, 3 = medium-dark, 4 = dark.

Spine ConditionCategorical: intact, partially broken, broken

Condition of the two spines. Intact means both are good, partially broken means one is worn or broken, and broken means both are worn or broken.

Spine Condition (numerical)

The spine condition coded as a number: 1 = both spines good, 2 = one spine worn or broken, 3 = both spines worn or broken.

Try it in the Regression module

• Logistic Regression: Fit a logistic regression model to visualize the probability of having satellites in terms of weight (or width). Estimate the effect of weight (or width) of the female crab on the odds of having satellites. Predict the probability of having satellites for a female crab that weighs 2kg. Find a confidence interval for this prediction.

Variables

Crab

ID of the female crab included in the study.

Number of SatellitesQuantitative

Number of male horseshoe crabs (satellites) nesting nearby, ranging from 0 to 9.

Weight (kg)Quantitative

Weight of the female horseshoe crab in kg.

Width (cm)Quantitative

Width of the shell of the female horseshoe crab in cm.

Shell Color

The color of the shell of the female horseshoe crab, acting as a proxy for age. The darker the color, the older the crab.

Shell Color (numerical)

The shell color coded as a number: 1 = medium-light, 2 = medium, 3 = medium-dark, 4 = dark.

Spine ConditionCategorical: intact, partially broken, broken

Condition of the two spines. Intact means both are good, partially broken means one is worn or broken, and broken means both are worn or broken.

Spine Condition (numerical)

The spine condition coded as a number: 1 = both spines good, 2 = one spine worn or broken, 3 = both spines worn or broken.

Try it in the Regression module

• Poisson Regression: Fit a logistic regression model to visualize the probability of having satellites in terms of weight (or width). Estimate the effect of weight (or width) of the female crab on the odds of having satellites. Predict the probability of having satellites for a female crab that weighs 2kg. Find a confidence interval for this prediction.

Variables

House

Identification number of the home, from 1 to 200.

Selling PriceQuantitative

Price the home sold for, in dollars.

Size (sqft)Quantitative

Floor area of the house, in square feet.

Lot Size (sqft)Quantitative

Area of the lot the house sits on, in square feet. An acre is 43,560 square feet, so the largest lots here run to several acres.

BedroomsQuantitative

Number of bedrooms.

BathsQuantitative

Number of bathrooms, where a half counts as 0.5 because it has no shower or tub.

AgeQuantitative

Age of the house at the time of sale, in years, ranging from 1 to 107.

GarageCategorical: Yes, No

Whether the home has a garage (Yes or No).

Age CategoryCategorical: New, Medium, Old

The same age grouped into three levels, New for 1 to 15 years, Medium for 16 to 50 years, and Old for 51 years and above.

Variables

Flower (Iris)

Identification number of the flower, from 1 to 150.

Type of IrisCategorical: Setosa, Versicolor, Virginica

The species (Setosa, Versicolor, or Virginica), with 50 flowers of each.

Sepal LengthQuantitative

Length (in cm) of the sepal, the green leaf-like part that encloses the bud before it opens.

Sepal WidthQuantitative

Width (in cm) of the sepal.

Petal LengthQuantitative

Length (in cm) of the petal, the colored part of the flower.

Petal WidthQuantitative

Width (in cm) of the petal.

Variables

WeekQuantitative

Week of experiment.

Week (since week 20)Quantitative

Number of weeks since week 20 of the experiment.

Weight (kg)Quantitative

Weight of leaf litter in the bag (in kg).

Try it in the Regression module

• Exponential Regression: Analyze the relationship between the number of weeks since week 20 (x-variable) and the weight of the bag (y-variable), and see if the litter decayed exponentially.

Variables

ChipCategorical: 7 categories

Name of computer chip.

YearQuantitative

Year chip was produced.

Year (since 1994)Quantitative

Number of years since 1994.

CountQuantitative

Number of components on the chip.

Try it in the Regression module

• Exponential Regression: Use Year (since 1994) to see whether Count still doubles each year.

Variables

Model

Make and model of the car.

Miles per gallonQuantitative

Fuel efficiency, in miles per gallon.

Weight (in 1000 lbs)Quantitative

Weight of the car, in thousands of pounds, so 2.62 means 2,620 pounds.

CylindersQuantitative

Number of cylinders in the engine (4, 6 or 8).

Displacement (cu.in.)Quantitative

The total volume swept by all the pistons, in cubic inches, which is the usual measure of engine size.

HorsepowerQuantitative

Gross horsepower of the engine.

TransmissionCategorical: Automatic, Manual

Whether the car has an Automatic or a Manual transmission.

Engine TypeCategorical: V-shaped, Straight

How the cylinders are arranged, either V-shaped in two angled banks or Straight in a single line.

GearsQuantitative

Number of forward gears (3, 4 or 5).

CarburetorsQuantitative

Number of carburetor barrels feeding the engine.

Variables

Loan Nr.

Identifier for the loan.

Loan AmountQuantitative, in USD

In USD, between $1,000 and $40,000.

DurationCategorical: 36, 60 months

Whether this is a loan to be paid back over 36 months or 60 months.

Interest RateQuantitative, a percentage

Quoted as a percentage.

GradeCategorical: A, B, C, D, E, F

A grade (from A to F) that lendingclub.com assigned to the loan, reflecting its risk. The grade of A indicates a loan with the smallest risk, while a grade of F indicates a loan with the greatest risk of defaulting.

HomeCategorical: own, mortgage, rent

Whether the person applying for the loan lives in a home without paying any mortgage ("own"), pays a mortgage ("mortgage"), or pays rent ("rent").

Income (in $1000)Quantitative, in 1,000 USD

Annual income (in 1,000 USD) of person applying for the loan.

FICO ScoreQuantitative, 300 to 850

Lower bound on the FICO score for the person taking out the loan. The FICO score is a three-digit number ranging from 300 to 850. It expresses how likely a person is to repay borrowed money based on their credit history.

Try it in the Regression module

• Simple Linear Regression: Select FICO Score as the x-Variable and Interest Rate as the y-Variable. The second tab draws the scatterplot with the fitted regression line and reports the correlation, r = -0.46: applicants with stronger credit scores are quoted lower rates. The third tab lists the fitted values and residuals, so you can find the loans the line misses by the most. Income as x and Loan Amount as y is the other strong pair, r = 0.51.
• Multiple Linear Regression: Select Interest Rate as the Response Variable and, under Select Explanatory Variables, FICO Score and Duration. The two are essentially uncorrelated with each other (r = 0.01) while each is related to the rate, so they contribute nearly separately to the fit. Add Home as a categorical explanatory variable to see it enter as dummy variables, and turn on Add Interactions to let the effect of FICO Score differ by loan duration. Leave Grade out: the platform assigns the grade from the risk of the loan, and mean interest rate climbs from 7.1% for grade A to 29.8% for grade F, so including it leaks the response into the predictors.

Variables

Penguin

Identification number of the penguin, from 1 to 342.

SpeciesCategorical: Adelie, Chinstrap, Gentoo

Penguin species (Adélie, Chinstrap, or Gentoo).

IslandCategorical: Dream, Torgersen, Biscoe

Island name (Dream, Torgersen, or Biscoe) in the Palmer Archipelago (Antarctica).

Bill Length (mm)Quantitative

Length (in mm) of a penguin's bill, measured along the upper ridge.

Bill Depth (mm)Quantitative

Depth (in mm) of a penguin's bill, measured from the top of the bill to the bottom of the bill.

Flipper Length (mm)Quantitative

Length (in mm) of the penguin's flipper.

Body Mass (g)Quantitative

Body mass (in g) of the penguin.

SexCategorical: Female, Male

Penguin sex (Female or Male). Nine penguins have a missing value.

Variables

State

The US state.

RegionCategorical: South, West, Northeast, Midwest

The region in the US that the state belongs to.

Student ExpenditureQuantitative

Current expenditure per pupil in fall enrollment for public elementary and secondary schools, for the 2012-13 school year.

SAT ScoreQuantitative

Average 2016 SAT score of all students in the state taking the SAT.

Participation PercentQuantitative

Percent of all students in a state who took the SAT in 2016.

Participation CategoryCategorical: low, medium, high

Participation percent classified as low, medium, or high, using cutoffs of 21% and 50%.

Adjusted Teacher SalaryQuantitative

Average teacher salary, adjusted for the cost of living in the state.

Teacher SalaryQuantitative

Mean annual salary of public elementary and secondary classroom teachers in the state, for the 2015-16 school year.

Student Teacher RatioQuantitative

Fall enrollment divided by the number of public elementary and secondary classroom teachers, in fall 2014.

Verbal SAT ScoreQuantitative

Average score on the verbal component of the 2016 SAT.

Math SAT ScoreQuantitative

Average score on the math component of the 2016 SAT.

Try it in the Regression module

• Simple Linear Regression: An apparent negative relationship between teacher salary (or student expenditure) and SAT score evaporates once we adjust for student participation in the SAT. You can visualize this by selecting "Participation Category" as the variable for coloring the points.

Variables

Student

Identification number of the student, from 1 to 171.

GenderCategorical: Female, Male

Female, male

AthleteCategorical: No, Yes

Whether the student is participating in college athletics

Self Esteem AppearanceQuantitative

A composite score on how much self esteem the student derives from their appearance. Obtained through a battery of questions of the form "When I think I look attractive, I feel good about myself." Each question was scored on a likert scale from 1 to 7 (1=strongly disagree, 7=strongly agree), with higher scores indicating higher self esteem, and the score represents the average over the battery of questions.

Self Esteem ApprovalQuantitative

A composite score on how much self esteem the student derives from how others approve of them. Obtained through a battery of questions of the form "I can't respect myself if others don't respect me." Each question was scored on a likert scale from 1 to 7 (1=strongly disagree, 7=strongly agree), with higher scores indicating higher self esteem, and the score represents the average over the battery of questions.

Eating DisorderQuantitative

A composite score measuring signs of mild eating disorders, obtained through a battery of 9 eating behavior questions such as "I throw up after I have eaten." All questions were scored from 1 to 6 (1="never", 2= "rarely", 3= "sometimes", 4="often", 5="usually", 6="always"), and the average score is reported.

Exercise FrequencyQuantitative

Hours per week exercising. Answer to the question "In the past month, on average, how many hours per week did you engage in physical exercise?"

Variables

Ship

ID of the ship included in the study.

Number of DamagesQuantitative

Number of damage incidents for the ship, ranging from 0 to 58.

Months in ServiceQuantitative

Number of months the ship has been in service.

Ship TypeCategorical: A, B, C, D, E

Type of the ship (A, B, C, D, or E).

Try it in the Regression module

• Poisson Regression: Fit a Poisson regression model to visualize the relationship between months in service (and/or ship type) and the number of damages. Estimate the effect of months in service on the expected number of damages. Predict the expected number of damages for a ship with a given number of months in service. Find a confidence interval for this prediction.

Variables

Flight

Flight number of the space shuttle program. The Challenger flight was the one right after STS-61-C.

Launch TemperatureQuantitative

Temperature, in degrees Fahrenheit, just before launch of the shuttle at Kennedy Space Center, Florida.

O-Ring DamageCategorical: No, Yes

Whether at least one of the O-rings in the two solid booster rockets got damaged during liftoff.

Try it in the Regression module

• Logistic Regression: One major concern of NASA engineers was that O-rings, made out of rubber, may not perform well when the launch temperature is low. Use the data on the previous space shuttle flights to model and visualize how O-ring failure depends on launch temperature. Is there evidence from these data that the launch temperature might have an effect on the probability of O-ring failure? The launch temperature for the Challenger flight was predicted to be just 31° F. Predict the probability of O-ring failure at such a low temperature. Are there any issues with your prediction?

Variables

Student

ID of the student included in the study.

Number of AwardsQuantitative

Number of awards received by the student, ranging from 0 to 6.

Math ScoreQuantitative

Math test score of the student.

Try it in the Regression module

• Poisson Regression: Fit a Poisson regression model to visualize the relationship between math score and the number of awards. Estimate the effect of math score on the expected number of awards. Predict the expected number of awards for a student with a given math score. Find a confidence interval for this prediction.

Variables

YearQuantitative

Year.

Decade (since 1900)Quantitative

Number of decades since 1900.

Population Size (Million)Quantitative

Size of U.S. population in millions.

Variables

Car

Identification number of the car, from 1 to 19.

Price ($)Quantitative

Asking price of the car, in U.S. dollars.

Age (years)Quantitative

Age of the car, in years.

HorsepowerQuantitative

Power of the engine, ranging from 38 to 130 horsepower.

TypeCategorical: US, Foreign

Where the car was manufactured, either US or Foreign. Ten of the cars are foreign built and nine are American.

Variables

Wine

Identification number of the wine, from 1 to 178.

Cultivar TypeCategorical: Barolo, Grignolino, Barbera

The grape variety the wine was made from (Barolo, Grignolino, or Barbera).

AlcoholQuantitative

Alcohol content, as a percentage.

Malic AcidQuantitative

Amount of malic acid, the sharp-tasting acid also found in green apples.

AshQuantitative

The mineral residue left behind after the wine is burned away.

AlcalinityQuantitative

Alcalinity of that ash, meaning how much acid is needed to neutralize it.

MagnesiumQuantitative

Amount of magnesium.

PhenolsQuantitative

Total phenols, the compounds that give wine much of its bitterness, color and astringency.

FlavanoidsQuantitative

The flavanoid phenols, which make up the largest group of phenols in wine.

NonflavanoidsQuantitative

The remaining, nonflavanoid phenols.

ProanthocyaninsQuantitative

Amount of proanthocyanins, a class of tannins that affect color and mouthfeel.

Color IntensityQuantitative

How deeply colored the wine is.

HueQuantitative

The shade of that color.

AbsorbanceQuantitative

The ratio of light absorbed at wavelengths of 280 and 315 nanometers, reported in the original source as OD280/OD315 of diluted wines.

ProlineQuantitative

Amount of proline, the most abundant amino acid in wine.

Variables

Animal

Identification number of the animal, from 1 to 24.

FamilyCategorical: Canidae, Felidae, Mustelidae

The family name of the animal (Canidae, Felidae, Mustelidae).

Common Name

The name under which the animal is commonly known.

Gestation (days)Quantitative

Average duration of gestational period, in days.

Longevity (yrs)Quantitative

Maximal life span, in years.

Birth weight (g)Quantitative

Average birth weight, in grams.

Adult weight (kg)Quantitative

Average adult weight, in kilograms.

Growth rate (per day)Quantitative

Postnatal growth-rate coefficient, in units of 1/day, from a Gompertz curve. A coefficient of 0.01 means it takes about 310 days to grow from 10% to 90% of adult weight, and that, at its fastest, the animal gains about 0.37% of its adult weight each day. That percentage is the coefficient divided by e. In this dataset the lowest value is the cougar, 0.0061: about 510 days to grow from 10% to 90% of adult weight, and about 0.22% of adult weight per day at its fastest (about 140 g a day for a 63 kg adult). The highest value is the least weasel, 0.0546: about 56 days, and about 2% of adult weight per day at its fastest (about 10 g a day for a 0.5 kg adult).

Variables

Car

Make and model of the vehicle.

Miles per GallonQuantitative

Fuel efficiency, in miles per gallon.

CylindersQuantitative

Number of cylinders: 3, 4, 5, 6, or 8.

Displacement (cu. in.)Quantitative

Engine displacement, in cubic inches.

HorsepowerQuantitative

Engine horsepower.

Weight (lbs)Quantitative

Vehicle weight, in pounds.

Acceleration (sec)Quantitative

Time to accelerate from 0 to 60 mph, in seconds.

Model YearQuantitative

Model year, given as the last two digits (70 = 1970).

OriginCategorical: American, European, Japanese

Where the car was built (American, European, or Japanese). The original data codes these as 1, 2 and 3.

Try it in the Machine Learning module

• Multiple Linear Regression: Use Horsepower and Weight (lbs) to predict Miles per Gallon.

Variables

Player

Player name.

Times at BatQuantitative

Number of times at bat in 1986.

HitsQuantitative

Number of hits in 1986.

Home RunsQuantitative

Number of home runs in 1986.

RunsQuantitative

Number of runs in 1986.

Runs Batted InQuantitative

Number of runs batted in in 1986.

WalksQuantitative

Number of walks in 1986.

Years in Major LeaguesQuantitative

Number of years in the major leagues.

Career Times at BatQuantitative

Number of times at bat during his career.

Career HitsQuantitative

Number of hits during his career.

Career Home RunsQuantitative

Number of home runs during his career.

Career RunsQuantitative

Number of runs during his career.

Career Runs Batted InQuantitative

Number of runs batted in during his career.

Career WalksQuantitative

Number of walks during his career.

LeagueCategorical: American, National

League at the end of 1986 (American or National).

DivisionCategorical: East, West

Division at the end of 1986 (East or West).

Put OutsQuantitative

Number of put outs in 1986.

AssistsQuantitative

Number of assists in 1986.

ErrorsQuantitative

Number of errors in 1986.

Salary ($1000s)Quantitative

1987 annual salary on opening day, in thousands of dollars.

League in 1987Categorical: American, National

League at the beginning of 1987 (American or National).

Try it in the Machine Learning module

• Multiple Linear Regression: Use Hits and Years in Major Leagues to predict Salary ($1000s).

Variables

Tract

Census tract number, used only as an identifier.

Crime Rate (per capita)Quantitative

Per capita crime rate by town.

Large Lot Zoning (%)Quantitative

Percent of residential land zoned for lots over 25,000 square feet.

Nonretail Business (%)Quantitative

Percent of non-retail business acres per town.

Charles RiverCategorical: Yes, No

Whether the tract bounds the Charles River (Yes or No).

Nitrogen Oxides (parts per 10 million)Quantitative

Nitrogen oxides concentration.

Rooms per DwellingQuantitative

Average number of rooms per dwelling.

Built Before 1940 (%)Quantitative

Percent of owner-occupied units built before 1940.

Distance to Employment CentersQuantitative

Weighted mean of distances to five Boston employment centers.

Highway Access IndexQuantitative

Index of accessibility to radial highways.

Property Tax Rate (per $10,000)Quantitative

Full-value property tax rate.

Pupil-Teacher RatioQuantitative

Pupil-teacher ratio by town.

Lower Status (%)Quantitative

Percentage of the population with lower socioeconomic status.

Median Home Value ($1000s)Quantitative

Median value of owner-occupied homes, in thousands of dollars.

Try it in the Machine Learning module

• Multiple Linear Regression: Use Lower Status (%) and Rooms per Dwelling to predict Median Home Value.

Variables

Store

Store number, used only as an identifier.

Unit Sales (1000s)Quantitative

Unit sales at each location, in thousands.

Competitor Price ($)Quantitative

Price charged by the competitor at each location.

Community Income ($1000s)Quantitative

Community income level, in thousands of dollars.

Advertising Budget ($1000s)Quantitative

Local advertising budget for the company, in thousands of dollars.

Population (1000s)Quantitative

Population size in the region, in thousands.

Price ($)Quantitative

Price the company charges for car seats at each site.

Shelf LocationCategorical: Bad, Medium, Good

Quality of the shelving location in the store (Bad, Medium, or Good).

Average AgeQuantitative

Average age of the local population.

Education LevelQuantitative

Education level at each location.

UrbanCategorical: Yes, No

Whether the store is in an urban location (Yes or No).

In USCategorical: Yes, No

Whether the store is in the US (Yes or No).

Try it in the Machine Learning module

• Multiple Linear Regression: Use Price ($) and Shelf Location to predict Unit Sales (1000s).

Variables

Id

ID of study participant.

AgeQuantitative

Age of participant in years.

SexCategorical: Female, Male

Sex of participant (Female/Male).

Yawn

Whether the participant yawned at least one time when watching the video (Yes or No).

Variables

Customer

Identification number of the customer, from 1 to 10,000.

DefaultCategorical: Yes, No

Whether the customer failed to repay their credit card debt (Yes for 333 customers, No for the other 9,667).

BalanceQuantitative

Average balance left on the credit card after the monthly payment is made, in dollars.

IncomeQuantitative

Income of the customer, in dollars.

StudentCategorical: Yes, No

Whether the customer is a student (Yes or No).

Try it in the Machine Learning module

• Multiple Logistic Regression: Use Balance and Student to predict Default.

Variables

Athlete

Identification number of the athlete, from 1 to 64.

AgeQuantitative

Age of the athlete, in years.

Weight (lbs)Quantitative

Total body weight, in pounds.

Height (in)Quantitative

Height, in inches.

Body Fat (%)Quantitative

Percentage of body weight that is fat.

Lean Body MassQuantitative

Body weight excluding fat, in pounds. This is computed from the two previous variables as weight times one minus the body fat proportion.

Repetitions 55Quantitative

Number of bench press repetitions completed before fatigue at a weight of 55 pounds (25 kg).

Repetitions 70Quantitative

Number of bench press repetitions completed before fatigue at a weight of 70 pounds (31.8 kg).

Max Bench PressQuantitative

The heaviest weight, in pounds, that the athlete could bench press a single time.

Variables

Patient

Identification number of the patient, from 1 to 303.

Heart DiseaseCategorical: Yes, No

Whether the patient was found to have heart disease, meaning a major artery narrowed by more than 50% (Yes for 139 patients, No for the other 164).

AgeQuantitative

Age of the patient, in years.

SexCategorical: Male, Female

Male or Female.

Chest Pain TypeCategorical: Typical Angina, Atypical Angina, Non-Anginal Pain, Asymptomatic

The kind of chest pain reported. Typical Angina is the classic pattern brought on by exertion, Atypical Angina fits the pattern only partly, Non-Anginal Pain is chest pain from some other cause, and Asymptomatic means the patient reported no chest pain at all.

Blood PressureQuantitative

Resting blood pressure on admission to the hospital, in mm Hg.

CholesterolQuantitative

Serum cholesterol, in mg/dl.

Blood SugarCategorical: Elevated, Regular

Fasting blood sugar, recorded as Elevated when it exceeds 120 mg/dl and Regular otherwise.

Maximum Heart RateQuantitative

The highest heart rate the patient reached during the exercise test, in beats per minute.

Exercise Induced AnginaCategorical: Yes, No

Whether the exercise itself brought on chest pain (Yes or No).

Old PeakQuantitative

How far the ST segment of the electrocardiogram was depressed by exercise, compared with the patient at rest. Larger values point to less blood reaching the heart muscle.

SlopeCategorical: Upsloping, Flat, Downsloping

The direction the ST segment took at peak exercise, either Upsloping, Flat or Downsloping.

Vessels ColoredQuantitative

How many of the major blood vessels, from 0 to 3, showed up as narrowed when dye was injected and viewed under X-ray.

Thallium ScanCategorical: Normal, Fixed Defect, Reversible Defect

Result of a thallium scan, which traces how well blood reaches the heart muscle. Normal means an even supply, a Fixed Defect is an area that stays poorly supplied even at rest, and a Reversible Defect is an area that runs short of blood during exercise but recovers at rest.

Try it in the Machine Learning module

• Multiple Logistic Regression: Use Chest Pain Type and Maximum Heart Rate to predict Heart Disease.

Variables

Crab

ID of the female crab included in the study.

Satellite

Whether the female crab had any male horseshoe crabs (satellites) nesting nearby (yes or no).

Weight (kg)Quantitative

Weight of the female horseshoe crab in kg.

Width (cm)Quantitative

Width of the shell of the female horseshoe crab in cm.

Shell Color

The color of the shell of the female horseshoe crab, acting as a proxy for age. The darker the color, the older the crab.

Shell Color (numerical)

The shell color coded as a number: 1 = medium-light, 2 = medium, 3 = medium-dark, 4 = dark.

Spine ConditionCategorical: intact, partially broken, broken

Condition of the two spines. Intact means both are good, partially broken means one is worn or broken, and broken means both are worn or broken.

Spine Condition (numerical)

The spine condition coded as a number: 1 = both spines good, 2 = one spine worn or broken, 3 = both spines worn or broken.

Variables

House

Identification number of the home, from 1 to 200.

Selling PriceQuantitative

Price the home sold for, in dollars.

Size (sqft)Quantitative

Floor area of the house, in square feet.

Lot Size (sqft)Quantitative

Area of the lot the house sits on, in square feet. An acre is 43,560 square feet, so the largest lots here run to several acres.

BedroomsQuantitative

Number of bedrooms.

BathsQuantitative

Number of bathrooms, where a half counts as 0.5 because it has no shower or tub.

AgeQuantitative

Age of the house at the time of sale, in years, ranging from 1 to 107.

GarageCategorical: Yes, No

Whether the home has a garage (Yes or No).

Age CategoryCategorical: New, Medium, Old

The same age grouped into three levels, New for 1 to 15 years, Medium for 16 to 50 years, and Old for 51 years and above.

Variables

Flower (Iris)

Identification number of the flower, from 1 to 150.

Type of IrisCategorical: Setosa, Versicolor, Virginica

The species (Setosa, Versicolor, or Virginica), with 50 flowers of each.

Sepal LengthQuantitative

Length (in cm) of the sepal, the green leaf-like part that encloses the bud before it opens.

Sepal WidthQuantitative

Width (in cm) of the sepal.

Petal LengthQuantitative

Length (in cm) of the petal, the colored part of the flower.

Petal WidthQuantitative

Width (in cm) of the petal.

Variables

Worker

Worker number, used only as an identifier.

YearQuantitative

Year the wage information was recorded.

AgeQuantitative

Age of the worker, in years.

Marital StatusCategorical: Never Married, Married, Widowed, Divorced, Separated

Never Married, Married, Widowed, Divorced, or Separated.

RaceCategorical: White, Black, Asian, Other

White, Black, Asian, or Other.

EducationCategorical: 1. < HS Grad, 2. HS Grad, 3. Some College, 4. College Grad, 5. Advanced Degree

Highest level completed, numbered so that the five levels stay in order: 1. < HS Grad, 2. HS Grad, 3. Some College, 4. College Grad, 5. Advanced Degree.

Job ClassCategorical: Industrial, Information

Industrial or Information.

HealthCategorical: <=Good, >=Very Good

Self-reported health level, <=Good or >=Very Good.

Health InsuranceCategorical: Yes, No

Whether the worker has health insurance (Yes or No).

Wage ($1000s)Quantitative

The worker's raw wage, in thousands of dollars per year.

Try it in the Machine Learning module

• Multiple Linear Regression: Use Age and Education to predict Wage ($1000s).

Variables

Model

Make and model of the car.

Miles per gallonQuantitative

Fuel efficiency, in miles per gallon.

Weight (in 1000 lbs)Quantitative

Weight of the car, in thousands of pounds, so 2.62 means 2,620 pounds.

CylindersQuantitative

Number of cylinders in the engine (4, 6 or 8).

Displacement (cu.in.)Quantitative

The total volume swept by all the pistons, in cubic inches, which is the usual measure of engine size.

HorsepowerQuantitative

Gross horsepower of the engine.

TransmissionCategorical: Automatic, Manual

Whether the car has an Automatic or a Manual transmission.

Engine TypeCategorical: V-shaped, Straight

How the cylinders are arranged, either V-shaped in two angled banks or Straight in a single line.

GearsQuantitative

Number of forward gears (3, 4 or 5).

CarburetorsQuantitative

Number of carburetor barrels feeding the engine.

Variables

Loan Nr.

Identifier for the loan.

Loan AmountQuantitative, in USD

In USD, between $1,000 and $40,000.

DurationCategorical: 36, 60 months

Whether this is a loan to be paid back over 36 months or 60 months.

Interest RateQuantitative, a percentage

Quoted as a percentage.

GradeCategorical: A, B, C, D, E, F

A grade (from A to F) that lendingclub.com assigned to the loan, reflecting its risk. The grade of A indicates a loan with the smallest risk, while a grade of F indicates a loan with the greatest risk of defaulting.

HomeCategorical: own, mortgage, rent

Whether the person applying for the loan lives in a home without paying any mortgage ("own"), pays a mortgage ("mortgage"), or pays rent ("rent").

Income (in $1000)Quantitative, in 1,000 USD

Annual income (in 1,000 USD) of person applying for the loan.

FICO ScoreQuantitative, 300 to 850

Lower bound on the FICO score for the person taking out the loan. The FICO score is a three-digit number ranging from 300 to 850. It expresses how likely a person is to repay borrowed money based on their credit history.

Try it in the Machine Learning module

• Multiple Linear Regression: Select Interest Rate as the Response Variable and FICO Score and Duration under Select Explanatory Variables, the same model as in the Regression module, then turn on Split into Train and Test Data to see how it holds up out of sample. The second tab shows the fitted model and plots, the third the fitted values and residuals. Grade does not belong in this model: the platform sets it from the risk of the loan, so it carries the answer with it.
• K-Means Clustering: Under Select Features choose Loan Amount and Income, the most strongly related pair in the dataset (r = 0.51), and turn on Standardize Features. Set Number of Clusters on the second tab and watch the centroids move as the algorithm iterates. Then use Color Dots with Grade: the clusters divide borrowers by how much they earn and borrow, which is not the same thing as how the platform grades the risk of the loan.
• K-Nearest Neighbors (Regression): Select Interest Rate as the Response Variable (Target) and FICO Score as the Feature, with Standardize Features and Train/Test Split on. The second tab draws the fit for the Number of Neighbors (k) you choose: a small k chases individual loans, while a large k flattens toward the overall mean rate of 12.8%. The third tab reports test-set performance, so you can compare several values of k.
• Principal Component Analysis: Under Select Features choose the four quantitative variables, Loan Amount, Interest Rate, Income and FICO Score, and turn on Standardize Features, since they are measured on very different scales. Turn on Show Percent Explained on the second tab to see how the variance splits, and read the loadings on the third tab. This dataset is a useful contrast with the ones where a single component dominates: the four variables form two weakly related pairs, Loan Amount with Income and Interest Rate with FICO Score, so PC1 takes 39.2% of the variance and PC2 another 35.3%, and it takes both to reach 74.5%.

Variables

Penguin

Identification number of the penguin, from 1 to 342.

SpeciesCategorical: Adelie, Chinstrap, Gentoo

Penguin species (Adélie, Chinstrap, or Gentoo).

IslandCategorical: Dream, Torgersen, Biscoe

Island name (Dream, Torgersen, or Biscoe) in the Palmer Archipelago (Antarctica).

Bill Length (mm)Quantitative

Length (in mm) of a penguin's bill, measured along the upper ridge.

Bill Depth (mm)Quantitative

Depth (in mm) of a penguin's bill, measured from the top of the bill to the bottom of the bill.

Flipper Length (mm)Quantitative

Length (in mm) of the penguin's flipper.

Body Mass (g)Quantitative

Body mass (in g) of the penguin.

SexCategorical: Female, Male

Penguin sex (Female or Male). Nine penguins have a missing value.

Variables

Day

Trading day number, used only as an identifier.

YearQuantitative

The year the observation was recorded.

Return 1 Day Ago (%)Quantitative

Percentage return on the previous trading day.

Return 2 Days Ago (%)Quantitative

Percentage return two trading days earlier.

Return 3 Days Ago (%)Quantitative

Percentage return three trading days earlier.

Return 4 Days Ago (%)Quantitative

Percentage return four trading days earlier.

Return 5 Days Ago (%)Quantitative

Percentage return five trading days earlier.

Volume (billions of shares)Quantitative

Number of shares traded that day, in billions.

Return Today (%)Quantitative

Percentage return on the day in question.

DirectionCategorical: Up, Down

Whether the market moved Up or Down on that day.

Try it in the Machine Learning module

• Multiple Logistic Regression: Use Return 1 Day Ago (%) and Volume (billions of shares) to predict Direction.

Variables

State

The US state.

RegionCategorical: South, West, Northeast, Midwest

The region in the US that the state belongs to.

Student ExpenditureQuantitative

Current expenditure per pupil in fall enrollment for public elementary and secondary schools, for the 2012-13 school year.

SAT ScoreQuantitative

Average 2016 SAT score of all students in the state taking the SAT.

Participation PercentQuantitative

Percent of all students in a state who took the SAT in 2016.

Participation CategoryCategorical: low, medium, high

Participation percent classified as low, medium, or high, using cutoffs of 21% and 50%.

Adjusted Teacher SalaryQuantitative

Average teacher salary, adjusted for the cost of living in the state.

Teacher SalaryQuantitative

Mean annual salary of public elementary and secondary classroom teachers in the state, for the 2015-16 school year.

Student Teacher RatioQuantitative

Fall enrollment divided by the number of public elementary and secondary classroom teachers, in fall 2014.

Verbal SAT ScoreQuantitative

Average score on the verbal component of the 2016 SAT.

Math SAT ScoreQuantitative

Average score on the math component of the 2016 SAT.

Variables

Student

Identification number of the student, from 1 to 171.

GenderCategorical: Female, Male

Female, male

AthleteCategorical: No, Yes

Whether the student is participating in college athletics

Self Esteem AppearanceQuantitative

A composite score on how much self esteem the student derives from their appearance. Obtained through a battery of questions of the form "When I think I look attractive, I feel good about myself." Each question was scored on a likert scale from 1 to 7 (1=strongly disagree, 7=strongly agree), with higher scores indicating higher self esteem, and the score represents the average over the battery of questions.

Self Esteem ApprovalQuantitative

A composite score on how much self esteem the student derives from how others approve of them. Obtained through a battery of questions of the form "I can't respect myself if others don't respect me." Each question was scored on a likert scale from 1 to 7 (1=strongly disagree, 7=strongly agree), with higher scores indicating higher self esteem, and the score represents the average over the battery of questions.

Eating DisorderQuantitative

A composite score measuring signs of mild eating disorders, obtained through a battery of 9 eating behavior questions such as "I throw up after I have eaten." All questions were scored from 1 to 6 (1="never", 2= "rarely", 3= "sometimes", 4="often", 5="usually", 6="always"), and the average score is reported.

Exercise FrequencyQuantitative

Hours per week exercising. Answer to the question "In the past month, on average, how many hours per week did you engage in physical exercise?"

Variables

State

Name of the US state.

Murder Arrests (per 100,000)Quantitative

Murder arrests per 100,000 residents.

Assault Arrests (per 100,000)Quantitative

Assault arrests per 100,000 residents.

Urban Population (%)Quantitative

Percent of the state population living in urban areas.

Rape Arrests (per 100,000)Quantitative

Rape arrests per 100,000 residents.

Variables

College

Name of the college or university.

PrivateCategorical: Yes, No

Whether the school is private rather than public (Yes or No).

Applications ReceivedQuantitative

Number of applications received.

Applications AcceptedQuantitative

Number of applications accepted.

New Students EnrolledQuantitative

Number of new students enrolled.

Top 10% of HS Class (%)Quantitative

Percent of new students from the top 10% of their high school class.

Top 25% of HS Class (%)Quantitative

Percent of new students from the top 25% of their high school class.

Full-time UndergraduatesQuantitative

Number of full-time undergraduates.

Part-time UndergraduatesQuantitative

Number of part-time undergraduates.

Out-of-State Tuition ($)Quantitative

Out-of-state tuition.

Room and Board ($)Quantitative

Room and board costs.

Book Costs ($)Quantitative

Estimated book costs.

Personal Spending ($)Quantitative

Estimated personal spending.

Faculty with PhD (%)Quantitative

Percent of faculty with a PhD.

Faculty with Terminal Degree (%)Quantitative

Percent of faculty with a terminal degree.

Student-Faculty RatioQuantitative

Number of students per faculty member.

Alumni Who Donate (%)Quantitative

Percent of alumni who donate.

Expenditure per Student ($)Quantitative

Instructional expenditure per student.

Graduation Rate (%)Quantitative

Graduation rate.

Variables

Car

Identification number of the car, from 1 to 19.

Price ($)Quantitative

Asking price of the car, in U.S. dollars.

Age (years)Quantitative

Age of the car, in years.

HorsepowerQuantitative

Power of the engine, ranging from 38 to 130 horsepower.

TypeCategorical: US, Foreign

Where the car was manufactured, either US or Foreign. Ten of the cars are foreign built and nine are American.

Variables

Wine

Identification number of the wine, from 1 to 178.

Cultivar TypeCategorical: Barolo, Grignolino, Barbera

The grape variety the wine was made from (Barolo, Grignolino, or Barbera).

AlcoholQuantitative

Alcohol content, as a percentage.

Malic AcidQuantitative

Amount of malic acid, the sharp-tasting acid also found in green apples.

AshQuantitative

The mineral residue left behind after the wine is burned away.

AlcalinityQuantitative

Alcalinity of that ash, meaning how much acid is needed to neutralize it.

MagnesiumQuantitative

Amount of magnesium.

PhenolsQuantitative

Total phenols, the compounds that give wine much of its bitterness, color and astringency.

FlavanoidsQuantitative

The flavanoid phenols, which make up the largest group of phenols in wine.

NonflavanoidsQuantitative

The remaining, nonflavanoid phenols.

ProanthocyaninsQuantitative

Amount of proanthocyanins, a class of tannins that affect color and mouthfeel.

Color IntensityQuantitative

How deeply colored the wine is.

HueQuantitative

The shade of that color.

AbsorbanceQuantitative

The ratio of light absorbed at wavelengths of 280 and 315 nanometers, reported in the original source as OD280/OD315 of diluted wines.

ProlineQuantitative

Amount of proline, the most abundant amino acid in wine.