Apply t test over all columns of data frame seperated by variable

Question

I have data frame like this

  X1 X2 X3 X4 X5 class
1  1  7  3  9  5     n
2  2  8  4 10  6     n
3  3  9  5  1  7     n
4  4 10  6  2  8     p
5  5  1  7  3  9     p
6  6  2  8  4 10     p

I like to run t test over all columns, separated on groups formed by the variable class. I know I can use for loop for this, but I wonder if there is better code for that.

m1 <- data.frame(matrix(c <- (1:10), nrow = 6, ncol = 5))

m1 <-  data.frame(m1,c("n","n","n","p","p","p"))
names(m1)[6] = "class"

# work for one column
t.test(X1~class, data= m1) 

# What I'm looking for
# t.test(X_i~class, data= m1)

`Map(function(x) t.test(reformulate('class', x), m1), names(m1)[1:5])` — rawr, Jul 24 '22 at 07:28
I found for loop easier than other methods :) `p_df <- m1 %>% filter(class == "p") n_df <- m1 %>% filter(class == "n") res <- vector() for (i in 1:(length(p_df)-1)){ res[i] <- t.test(p_df[,i],n_df[,i])$p.value }` — Amir Asaadi, Aug 01 '22 at 13:30

socialscientist · Answer 1 · 2022-07-24T06:29:14.467

library(dplyr)
library(tidyr)
library(broom)

df <- data.frame(
  x1 = 0:9,
  x2 = 10:19,
  x3 = 20:29,
  class = rep(c("a", "b"), each = 5)
)

# Conduct tests and store in nested data.frame
nested <- df %>%
  group_by(class) %>%
  summarise(across(everything(),  ~ list(
    t.test(.) %>%
      tidy() %>%
      select(estimate, statistic,
             p.value, conf.low, conf.high)
  )))

# Unnest and turn into long structure
long <- nested %>%
  unnest(cols = starts_with("x"), names_sep = "_") %>%
  pivot_longer(cols = starts_with("x"),
               names_to = "quantity",
               values_to = "value")


# Split variables into multiple columns and widen 
long %>%
  separate(col = quantity, into = c("variable", "quantity"), sep = "_") %>%
  pivot_wider(names_from = "quantity")
#> # A tibble: 6 x 7
#>   class variable estimate statistic    p.value conf.low conf.high
#>   <chr> <chr>       <dbl>     <dbl>      <dbl>    <dbl>     <dbl>
#> 1 a     x1              2      2.83 0.0474       0.0368      3.96
#> 2 a     x2             12     17.0  0.0000707   10.0        14.0 
#> 3 a     x3             22     31.1  0.00000636  20.0        24.0 
#> 4 b     x1              7      9.90 0.000584     5.04        8.96
#> 5 b     x2             17     24.0  0.0000178   15.0        19.0 
#> 6 b     x3             27     38.2  0.00000281  25.0        29.0

Apply t test over all columns of data frame seperated by variable

1 Answers1