get_summary_stats with grouped_data

[dhofstetter]

The get_summary_stats method is behaves different for grouped and ungrouped data. For grouped data, the user defined probabilites are ignored completely.

The example below shows the problem observed.

require(tidyverse)
require(assertthat)
require(rstatix)

# We need some probabilites we want to observe as quantiles
probabilities_to_observe <- c(0, .5, .66, 1)

# Create an arbitrary frame with a grouping variable (actually the groups are
# the same which is better to visualize the behavioral problem)
data <- data_frame(
    gr=as.ordered(c("G1", "G1", "G1", "G1", "G1")), 
    value=runif(5,0,10)
)

summary1 <- data %>% 
    group_by(gr) %>% 
    get_summary_stats(value, type="quantile", probs = probabilities_to_observe)

summary2 <- data %>%
    get_summary_stats(value, type="quantile", probs = probabilities_to_observe)

# Check for expected outputs
names2_to_have <- c("variable", "n", "0%", "50%", "66%", "100%")
names1_to_have <- c("gr", names2_to_have)
    
assertthat::are_equal(names(summary1), names1_to_have)
assertthat::are_equal(names(summary2), names2_to_have)

BR Daniel

[kassambara]

fixed now, thanks