CODEWITHSUNDEEP
rdplyr
Ahdee
Ahdee

Reputation: 4949

Getting value not found while running chisq_test using dplyr?

I have two factors, day and the other group. The value is how many totals in each group.

x <- c(5,17,31,9,17,10,30,28,16,29,14,34)
y <- c(1,2,3,4,5,6,7,8,9,10,11,12)
day <- as.character (c ( 1,2,3,4,5,6,7,8,9,10,11,12) )


df1 <- data.frame(x, y, day)
df2 <- reshape2::melt(df1, id.vars='day')
colnames (df2)[2] = "group"



> df2
   day group value
1    1     x     5
2    2     x    17
3    3     x    31
4    4     x     9
5    5     x    17
6    6     x    10
7    7     x    30
8    8     x    28
9    9     x    16
10  10     x    29
11  11     x    14
12  12     x    34
13   1     y     1
14   2     y     2
15   3     y     3
16   4     y     4
17   5     y     5
18   6     y     6
19   7     y     7
20   8     y     8
21   9     y     9
22  10     y    10
23  11     y    11
24  12     y    12

so in the above example there are a total of 5 in group x and total of 1 for y in day 1. I would like to determine significance for each unique days between x and y, so perhaps a chisquare test?

I run the code as such but for some reason, it keeps implying that the group is not found?

df2 %>% group_by(day) %>% 
  rstatix::chisq_test( day, group ) %>%
  add_significance("p.adj")

Can anyone help with this, thanks in advance.

Upvotes: 1

Views: 81

Answers (1)

akrun
akrun

Reputation: 887231

Perhaps this helps

library(dplyr)
library(purrr)
library(rstatix)
df2 %>%
   mutate(day = as.numeric(day)) %>%
   split(.$day) %>% 
   map_dfr(~  with(.x, chisq_test(setNames(value, group))), .id = 'day')

-output

# A tibble: 12 × 7
   day       n statistic          p    df method          p.signif
   <chr> <int>     <dbl>      <dbl> <dbl> <chr>           <chr>   
 1 1         2      2.67 0.102          1 Chi-square test ns      
 2 2         2     11.8  0.000579       1 Chi-square test ***     
 3 3         2     23.1  0.00000157     1 Chi-square test ****    
 4 4         2      1.92 0.166          1 Chi-square test ns      
 5 5         2      6.55 0.0105         1 Chi-square test *       
 6 6         2      1    0.317          1 Chi-square test ns      
 7 7         2     14.3  0.000156       1 Chi-square test ***     
 8 8         2     11.1  0.000858       1 Chi-square test ***     
 9 9         2      1.96 0.162          1 Chi-square test ns      
10 10        2      9.26 0.00235        1 Chi-square test **      
11 11        2      0.36 0.549          1 Chi-square test ns      
12 12        2     10.5  0.00118        1 Chi-square test **

Or could use group_modify

df2 %>% 
   group_by(day = as.numeric(day)) %>%
   group_modify(~ with(.x, chisq_test(setNames(value, group)))) %>% 
   ungroup
# A tibble: 12 × 7
     day     n statistic          p    df method          p.signif
   <dbl> <int>     <dbl>      <dbl> <dbl> <chr>           <chr>   
 1     1     2      2.67 0.102          1 Chi-square test ns      
 2     2     2     11.8  0.000579       1 Chi-square test ***     
 3     3     2     23.1  0.00000157     1 Chi-square test ****    
 4     4     2      1.92 0.166          1 Chi-square test ns      
 5     5     2      6.55 0.0105         1 Chi-square test *       
 6     6     2      1    0.317          1 Chi-square test ns      
 7     7     2     14.3  0.000156       1 Chi-square test ***     
 8     8     2     11.1  0.000858       1 Chi-square test ***     
 9     9     2      1.96 0.162          1 Chi-square test ns      
10    10     2      9.26 0.00235        1 Chi-square test **      
11    11     2      0.36 0.549          1 Chi-square test ns      
12    12     2     10.5  0.00118        1 Chi-square test **

Upvotes: 2

Related Questions