| Title: | Datasets from Computer Age Statistical Inference |
| Version: | 0.2.1 |
| Description: | Provides the datasets from Efron & Hastie (2016, ISBN: 9781108107952), "Computer Age Statistical Inference: Algorithms, Evidence, and Data Science", in an accessible R format for those who want to use them for study or to try to reproduce analyses from the book. |
| License: | GPL (≥ 3) |
| Encoding: | UTF-8 |
| Depends: | R (≥ 3.5.0) |
| Suggests: | knitr, rmarkdown, vcdExtra, car |
| VignetteBuilder: | knitr |
| RoxygenNote: | 7.3.3 |
| Language: | en-US |
| URL: | https://github.com/friendly/CASIdata, https://friendly.github.io/CASIdata/ |
| BugReports: | https://github.com/friendly/CASIdata/issues |
| NeedsCompilation: | no |
| Packaged: | 2026-02-10 01:24:21 UTC; friendly |
| Author: | Michael Friendly [aut, cre] |
| Maintainer: | Michael Friendly <friendly@yorku.ca> |
| Repository: | CRAN |
| Date/Publication: | 2026-02-12 08:00:20 UTC |
DTI Brain Imaging Data
Description
Diffusion Tensor Imaging (DTI) data comparing 6 dyslexic children with 6 normal controls, from Figures 15.9 and 15.10. Z scores were computed at 15,443 three-dimensional brain coordinates (voxels).
Format
A data frame with 15443 rows and 4 variables:
- x
Voxel coordinate: back to front
- y
Voxel coordinate: left to right
- z
Voxel coordinate: bottom to top
- Zscore
Z score comparing dyslexic vs normal controls at this voxel
Source
https://hastie.su.domains/CASI_files/DATA/DTI.csv
References
Efron, B. and Hastie, T. (2016). Computer Age Statistical Inference. Cambridge University Press, Figures 15.9 and 15.10.
Examples
data(DTI)
str(DTI)
ALS Data
Description
Data on amyotrophic lateral sclerosis (Lou Gehrig's disease) from Section 17.2. There are 1822 observations on individuals with ALS. The goal is to predict the rate of progression dFRS of a functional rating score, using 369 predictors based on measurements (and derivatives of these) obtained from patient visits.
Format
A data frame with 1822 rows and 371 variables. The key variables are
testset (logical indicator for training/test split) and dFRS
(response: rate of progression of the ALS functional rating score). The 369
predictor variables include:
Demographics:
Age,Sex.Male,Sex.Female, and race indicators (Race...Caucasian,Race...Asian, etc.)Family history of neurological diseases in relatives (e.g.,
Father,Mother,Brother,Sister)Neurological disease indicators (e.g.,
Neurological.Disease.ALS,Neurological.Disease.PARKINSON.S.DISEASE)Site of onset (
Site.of.Onset.Onset..Bulbar,Site.of.Onset.Onset..Limb)Symptoms (
Symptom.Atrophy,Symptom.Cramps,Symptom.Fasciculations,Symptom.Speech, etc.)Study arm indicators (
Study.Arm.ACTIVE,Study.Arm.PLACEBO)Clinical measurements with summary statistics (first, last, min, max, mean, sd, slope): ALSFRS scores, blood pressure, forced/slow vital capacity (
fvc.liters,svc.liters), respiratory rate, weight, heightALSFRS subscale items:
climbing.stairs,cutting,dressing,handwriting,salivation,speech,swallowing,turning,walking
Details
These data were kindly provided by Lester Mackey and Lilly Fang, who won the DREAM challenge prediction prize in 2012 (Kuffner et al., 2015). It includes some additional variables created by them. Their winning entry used Bayesian trees, not too different from random forests.
Source
https://hastie.su.domains/CASI_files/DATA/ALS.txt
References
Efron, B. and Hastie, T. (2016). Computer Age Statistical Inference. Cambridge University Press, Section 17.2.
Examples
data(als)
str(als)
Baseball Batting Averages
Description
Batting averages for 18 Major League players in the 1970 season, from Table 7.1. This dataset illustrates empirical Bayes estimation, where early-season performance is used to predict full-season batting averages.
Format
A data frame with 18 rows and 3 variables:
- Player
Player ID number
- MLE
Batting average based on the first 90 at-bats of the season
- TRUTH
Batting average for the remainder of the 1970 season
Source
https://hastie.su.domains/CASI_files/DATA/baseball.txt
References
Efron, B. and Hastie, T. (2016). Computer Age Statistical Inference. Cambridge University Press, Table 7.1.
Examples
data(baseball)
str(baseball)
Bivariate Normal Data
Description
40 points generated from a bivariate normal distribution, with some entries missing. From Figure 9.3.
Format
A data frame with 40 rows and 2 variables:
- X1
First variable
- X2
Second variable
Source
https://hastie.su.domains/CASI_files/DATA/bivnorm.csv
References
Efron, B. and Hastie, T. (2016). Computer Age Statistical Inference. Cambridge University Press, Figure 9.3.
Examples
data(bivnorm)
str(bivnorm)
Butterfly Species Data
Description
Number of butterfly species seen a given number of times each in two years of trapping. From Table 6.2. This is a frequency data frame.
Format
A data frame with 24 rows and 2 variables:
- k
Number of times a species was trapped
- count
Number of species seen exactly k times (e.g., 118 species trapped just once, 74 trapped twice each)
Source
https://hastie.su.domains/CASI_files/DATA/butterfly.txt
References
Efron, B. and Hastie, T. (2016). Computer Age Statistical Inference. Cambridge University Press, Table 6.2.
Examples
data(butterfly)
str(butterfly)
Cell Infusion Data
Description
Human cell colonies infused with mouse nuclei in 5 different ratios over 1 to 5 days. From Table 8.2.
Format
A data frame with 25 rows and 4 variables:
- thrived
Number of cells that thrived
- N
Colony size (number of cells)
- ratio
Ratio of mouse nuclei to human cells (1-5)
- time
Day of observation (1-5)
Source
https://hastie.su.domains/CASI_files/DATA/cellinfusion.txt
References
Efron, B. and Hastie, T. (2016). Computer Age Statistical Inference. Cambridge University Press, Table 8.2.
Examples
data(cellinfusion)
str(cellinfusion)
Cholesterol Data
Description
Cholestyramine, a proposed cholesterol lowering drug, was administered to 164 men for an average of seven years each. From Figure 20.1.
Format
A data frame with 164 rows and 2 variables:
- compliance
Fraction of intended dose actually taken (standardized)
- cholesterol.decrease
Decrease in cholesterol level over the course of the experiment
Source
https://hastie.su.domains/CASI_files/DATA/cholesterol.txt
References
Efron, B. and Hastie, T. (2016). Computer Age Statistical Inference. Cambridge University Press, Figure 20.1.
Examples
data(cholesterol)
str(cholesterol)
Diabetes Data
Description
Data from 442 diabetes patients used in Section 7.3. The response is a quantitative measure of disease progression one year after baseline. There are ten baseline predictors: age, sex, body-mass index, average blood pressure, and six blood serum measurements.
Format
A data frame with 442 rows and 12 variables:
- X
Row index
- age
Age of patient
- sex
Sex of patient
- bmi
Body mass index
- map
Average blood pressure (mean arterial pressure)
- tc
Total cholesterol (serum measurement)
- ldl
Low-density lipoproteins (serum measurement)
- hdl
High-density lipoproteins (serum measurement)
- tch
Total cholesterol / HDL (serum measurement)
- ltg
Log of triglycerides (serum measurement)
- glu
Blood sugar level (serum measurement)
- prog
Response: quantitative measure of disease progression
Details
First used in the LARS paper (Efron et al., 2004).
Note: In Table 7.2, the centered predictor variables were standardized to unit L2 norm. In Table 20.1 they were standardized to unit variance.
Source
https://hastie.su.domains/CASI_files/DATA/diabetes.csv
References
Efron, B., Hastie, T., Johnstone, I. and Tibshirani, R. (2004). Least Angle Regression. Annals of Statistics, 32(2), 407-499.
Efron, B. and Hastie, T. (2016). Computer Age Statistical Inference. Cambridge University Press, Section 7.3.
Examples
data(diabetes)
str(diabetes)
Dose Response Data
Description
Data from 11 groups of mice (10 each) exposed to drug Xilathon at different doses. From Figure 8.2.
Format
A data frame with 11 rows and 2 variables:
- Dose
Log dose level (each step is a doubling)
- Proportion
Proportion of mice that died at that dose
Source
https://hastie.su.domains/CASI_files/DATA/doseresponse.txt
References
Efron, B. and Hastie, T. (2016). Computer Age Statistical Inference. Cambridge University Press, Figure 8.2.
Examples
data(doseresponse)
str(doseresponse)
Galaxy Data
Description
Counts of galaxies binned by redshift and magnitude, from Table 8.5. The data have been reshaped into long format with variables for magnitude, redshift category, and frequency count.
Format
A data frame with 270 rows and 3 variables:
- mag
Magnitude category (1-18)
- red
Redshift category (1-15)
- freq
Number of galaxies in this bin
Source
https://hastie.su.domains/CASI_files/DATA/galaxy.txt
References
Efron, B. and Hastie, T. (2016). Computer Age Statistical Inference. Cambridge University Press, Table 8.5.
Examples
data(galaxy)
str(galaxy)
library(car)
## Fit a main effects Poisson GLM
# This treats `mag` and `red` as numeric
galaxy.mod0 <- glm(freq ~ mag + red,
data = galaxy, family = poisson)
Anova(galaxy.mod0)
## Fit response surface model
galaxy.mod1 <- glm(freq ~ poly(mag,2) +
poly(red, 2) +
mag : red,
data = galaxy, family = poisson)
Anova(galaxy.mod1)
summary(galaxy.mod1)
Human Ancestry Haplotype Data
Description
Genotype data for 197 US individuals from 4 racial groups (African American, European, Japanese, and African) at 100 SNP locations. From Section 13.5.
Format
A data frame with 197 rows and 102 variables. The first column
X is a row index, race is the racial/ethnic group identifier,
and the remaining 100 columns (Snp1 through Snp100) contain
genotype values (0, 1, or 2) at each SNP location, with NA for
missing values.
Source
https://hastie.su.domains/CASI_files/DATA/haplotype.csv
References
Efron, B. and Hastie, T. (2016). Computer Age Statistical Inference. Cambridge University Press, Section 13.5.
Examples
data(haplotype)
str(haplotype)
Insurance Life Table Data
Description
Insurance company life table from Table 9.1. At each age, gives the number of policy holders and the number of deaths.
Format
A data frame with rows for each age group and 3 variables:
- age
Age of policy holders
- n
Number of policy holders at this age
- y
Number of deaths at this age
Source
https://hastie.su.domains/CASI_files/DATA/insurance.txt
References
Efron, B. and Hastie, T. (2016). Computer Age Statistical Inference. Cambridge University Press, Table 9.1.
Examples
data(insurance)
str(insurance)
Leukemia Gene Expression Data (Small)
Description
Gene expression measurements on 72 leukemia patients: 47 ALL (acute lymphoblastic leukemia) and 25 AML (acute myeloid leukemia). From the landmark Golub et al. (1999) Science paper. This smaller subset contains 3571 genes and is used in Section 19.1.
Format
A data frame with 3571 rows (genes) and 72 columns (patients). Column names indicate the class label (ALL or AML) for each patient.
Details
A larger dataset with 7128 genes is also available from the CASI website.
Source
https://hastie.su.domains/CASI_files/DATA/leukemia_small.csv
References
Golub, T.R., et al. (1999). Molecular Classification of Cancer: Class Discovery and Class Prediction by Gene Expression Monitoring. Science, 286, 531-537.
Efron, B. and Hastie, T. (2016). Computer Age Statistical Inference. Cambridge University Press, Section 19.1.
Examples
data(leukemia_small)
str(leukemia_small)
NCOG Head and Neck Cancer Data
Description
Head and neck cancer survival data from the Northern California Oncology Group (NCOG), from Section 9.2. Patients were randomized to one of two treatment arms.
Format
A data frame with survival time information and variables:
- t
Time in months until death or censoring
- d
Death indicator: 1 = death observed, 0 = censored
- arm
Treatment arm: "A" = Chemotherapy, "B" = Chemotherapy + Radiation
- day
Day of event/censoring
- month
Month of event/censoring
- year
Year of event/censoring
Source
https://hastie.su.domains/CASI_files/DATA/ncog.txt
References
Efron, B. and Hastie, T. (2016). Computer Age Statistical Inference. Cambridge University Press, Section 9.2.
Examples
data(ncog)
str(ncog)
Lymph Nodes Cancer Data
Description
Data on lymph nodes removed from 844 cancer patients, from Figure 6.3.
Format
A data frame with 844 rows and 2 variables:
- n
Number of lymph nodes removed
- x
Number of nodes found to be positive (malignant)
Source
https://hastie.su.domains/CASI_files/DATA/nodes.txt
References
Efron, B. and Hastie, T. (2016). Computer Age Statistical Inference. Cambridge University Press, Figure 6.3.
Examples
data(nodes)
str(nodes)
Pediatric Cancer Survival Data
Description
Survival data on 1620 children with cancer, from Section 9.4 and Table 9.6.
Format
A data frame with 1620 rows and 7 variables:
- sex
Sex: 1 = male, 2 = female
- race
Race: 1 = white, 2 = nonwhite
- age
Age in years
- entry
Calendar date of entry in days since July 1, 2001
- far
Home distance from treatment center in miles
- t
Survival time in days
- d
Death indicator: 1 = death observed, 0 = censored
Source
https://hastie.su.domains/CASI_files/DATA/pediatric.txt
References
Efron, B. and Hastie, T. (2016). Computer Age Statistical Inference. Cambridge University Press, Section 9.4, Table 9.6.
Examples
data(pediatric)
str(pediatric)
Police Racial Bias Data
Description
Z scores for 2749 New York City police officers, from Figure 15.7. A large value suggests racial bias in policing behavior.
Format
A data frame with 2749 rows and 1 variable:
- z
Z score measuring potential racial bias
Source
https://hastie.su.domains/CASI_files/DATA/police.txt
References
Efron, B. and Hastie, T. (2016). Computer Age Statistical Inference. Cambridge University Press, Figure 15.7.
Examples
data(police)
str(police)
Prostate Cancer Z-values
Description
Vector of 6033 z-values comparing gene expression between prostate cancer patients and controls, as pictured in Figure 3.4. These were computed as described on page 272.
Format
A data frame with 6033 rows and 1 variable:
- z
Z-value for each gene comparing cancer vs control expression
Source
https://hastie.su.domains/CASI_files/DATA/prostz.txt
References
Efron, B. and Hastie, T. (2016). Computer Age Statistical Inference. Cambridge University Press, Section 3.3, Figure 3.4.
Examples
data(prostz)
str(prostz)
Student Score Data
Description
Test scores for 22 students on 5 different exams, from Tables 3.1 and 10.1.
Format
A data frame with 22 rows and 5 variables:
- mech
Mechanics exam score
- vecs
Vectors exam score
- alg
Algebra exam score
- analy
Analysis exam score
- stat
Statistics exam score
Source
https://hastie.su.domains/CASI_files/DATA/student_score.txt
References
Efron, B. and Hastie, T. (2016). Computer Age Statistical Inference. Cambridge University Press, Tables 3.1 and 10.1.
Examples
data(student_score)
str(student_score)
Type Ia Supernova Data
Description
Measurements from 39 Type Ia supernovas, from Figure 12.1 and Table 12.1. These supernovas were close enough to Earth to observe their actual magnitudes. The goal is to predict magnitude from spectral energy measurements.
Format
A data frame with 39 rows and 11 variables:
- Magnitude
Actual observed magnitude of the supernova
- E1
Spectral energy in frequency band 1
- E2
Spectral energy in frequency band 2
- E3
Spectral energy in frequency band 3
- E4
Spectral energy in frequency band 4
- E5
Spectral energy in frequency band 5
- E6
Spectral energy in frequency band 6
- E7
Spectral energy in frequency band 7
- E8
Spectral energy in frequency band 8
- E9
Spectral energy in frequency band 9
- E10
Spectral energy in frequency band 10
Source
https://hastie.su.domains/CASI_files/DATA/supernova.txt
References
Efron, B. and Hastie, T. (2016). Computer Age Statistical Inference. Cambridge University Press, Figure 12.1, Table 12.1.
Examples
data(supernova)
str(supernova)
Vasoconstriction Data
Description
Data on vasoconstriction (lung constriction) response, from Table 13.2.
Format
A data frame with 39 rows and 2 variables:
- volume
Volume measurement
- constriction
Logical: TRUE if constriction occurred, FALSE otherwise
Source
https://hastie.su.domains/CASI_files/DATA/vasoconstriction.txt
References
Efron, B. and Hastie, T. (2016). Computer Age Statistical Inference. Cambridge University Press, Table 13.2.
Examples
data(vasoconstriction)
str(vasoconstriction)