datarium: Data bank for statistical analyses and visualization

Contains 23 curated data sets for teaching classical statistical inference in R. Each data set is paired with the analysis it is designed to teach — one data set per test. The same map is available in R via ?datarium. Each analysis below links to its worked tutorial on Datanovia.

Comparing means

Analysis	Data set
One-sample t-test	mice
Paired-samples t-test	mice2
Two-samples (independent) t-test	genderweight

Analysis of variance (ANOVA)

Analysis	Data set
Two-way ANOVA	jobsatisfaction
Three-way ANOVA	headache, heartattack
One-way repeated measures ANOVA	selfesteem
Two-way repeated measures ANOVA	selfesteem2
Three-way repeated measures ANOVA	weightloss
Two-way mixed ANOVA	anxiety, depression
Three-way mixed ANOVA	performance
Two-way ANCOVA	stress

Categorical data and proportions

Analysis	Data set
Chi-square test of independence	properties, housetasks.raw
Categorical descriptive statistics	titanic.raw
McNemar’s test	antismoking
Cochran’s Q test	taskachievment
Cochran-Armitage trend test	renalstone

Regression

Analysis	Data set
Linear regression	marketing
Poisson regression (count data)	infections
Logistic regression (real, messy clinical data)	heartdisease

Time series

Analysis	Data set
Time series	AirPassengersDf

Helper functions

The two data sets (Titanic and housetasks) are frequency/contingency table. We’ll create our demo data sets by recovering the original data from Titanic and housetasks tables.

To do so, first copy and paste the following helper function:

counts_to_cases <- function(x, countcol = "Freq") {
    if(!inherits(x, "table")) x <- as.table(as.matrix(x))
      x <- as.data.frame(x)
    # Get the row indices to pull from x
    idx <- rep.int(seq_len(nrow(x)), x[[countcol]])
    # Drop count column
    x[[countcol]] <- NULL
    # Get the rows from x
    x <- x[idx, ]
    rownames(x) <- 1:nrow(x)
    x
}

Then, recover the original data as follow:

# Load the data
data("Titanic")
data("housetasks", package = "factoextra")

# Recover the original raw data
titanic.raw <- counts_to_cases(Titanic)
housetasks.raw <- counts_to_cases(housetasks)