cumulative mean by group with NA in r

Question

I wanted to use cummean by groups ignoring NA. Here is the sample data.

data.frame(group1 = c(rep(1,3),rep(2,3),rep(3,3)),
           val = c(c(NA,4,5),c(NA,NA,5),c(1,NA,3)))

How can I get cumulative mean by group ignoring NA?

Here is my desired out

c(NA,4,4.5,NA,NA,5,1,1,2)

I don't think the duplicate link address the question OP asked — akrun, Oct 24 '20 at 20:00
Here is more appropriate duplicate: https://stackoverflow.com/questions/49906503/using-cummean-with-group-by-and-ignoring-nas — missuse, Oct 24 '20 at 20:02

akrun · Answer 1 · 2020-10-24T20:09:14.747

2

We can use replace to ignore the NA elements in calculating the cummean

library(dplyr)
library(tidyr)
df1 %>%
     group_by(group1) %>%
     mutate(new = replace(val, !is.na(val), cummean(val[!is.na(val)]))) %>%
     fill(new)

-output

# A tibble: 9 x 3
# Groups:   group1 [3]
#  group1   val   new
#   <dbl> <dbl> <dbl>
#1      1    NA  NA  
#2      1     4   4  
#3      1     5   4.5
#4      2    NA  NA  
#5      2    NA  NA  
#6      2     5   5  
#7      3     1   1  
#8      3    NA   1  
#9      3     3   2

edited Oct 24 '20 at 20:09

answered Oct 24 '20 at 19:59

akrun

874,273
37
540
662

1

I ignore NAs not treat those as zero. Can you take a look at the desired output? – John legend2 Oct 24 '20 at 20:04
@Johnlegend2 i think in the expected output, you showed `5 1 1 2` instead of `5 1 NA 2` because the 8th obs is NA in input – akrun Oct 24 '20 at 20:07
Thank you so much. How can I have desired output? can I have non-NA value 8th row? – John legend2 Oct 24 '20 at 20:08
@Johnlegend2 you can use `fill` updated the post – akrun Oct 24 '20 at 20:09

cumulative mean by group with NA in r

1 Answers1