remove Rows with complete set of NA-CodePudding

I have a dataset and I would like to delete the rows that have a complete set of NAs in columns 456:555, I want to keep those with some NAs but I need to delete those with a complete set of NAs

I have tried

final[complete.cases(final[ , 456:555]),]

but this doesn't work. It says

Error in help.search(c("[", "final", "complete.cases(final[, c(456:555)])",  : argument ‘pattern’ must be a single character string

then I think this probably would work:

data[rowSums(is.na(data)) != ncol(data),]

but I don't know where to include 456:555 there

what should I do?

Thanks!

CodePudding user response：

Maybe you can do something like this, not the cleanest approach:



# data frame with one row complete NA
df <- data.frame(V1 = c(NA, 3, NA, 2, 3),
                 V2 = c(NA, 3, 1, NA, 5),
                 V3 = c(NA, NA, NA ,NA, NA))

df

V1 V2 V3
1 NA NA NA
2  3  3 NA
3 NA  1 NA
4  2 NA NA
5  3  5 NA



old_df <- df[4:5,] # get rows you wanna keep regardless of number of NAs
new_df <- df[1:3,] # get rows where you wanna delete complete NAs

# "delete" complete NAs
new_df <-  new_df %>% 
    filter(is.na(new_df) %>% rowSums() != length(new_df))

# build the two dfs together
df <-rbind(old_df, new_df)
df

V1 V2 V3
4  2 NA NA
5  3  5 NA
1  3  3 NA
2 NA  1 NA

CodePudding user response：

Here is one simple solution with the sjmisc package:

df[!apply(df[456:555],1,sjmisc::all_na),]

To check that it does what you want, please find a little reprex:

REPREX

df <- data.frame(V1 = c(NA, 3, NA, 2, 3),
                 V2 = c(NA, 3, 1, NA, 5),
                 V3 = c(NA, NA, NA ,NA, NA))

df
#>   V1 V2 V3
#> 1 NA NA NA
#> 2  3  3 NA
#> 3 NA  1 NA
#> 4  2 NA NA
#> 5  3  5 NA

# Select all line except `all_na` lines for the selected columns:
df[!apply(df[2:3],1,sjmisc::all_na),]
#>   V1 V2 V3
#> 2  3  3 NA
#> 3 NA  1 NA
#> 5  3  5 NA

^{Created on 2021-10-11 by the reprex package (v2.0.1)}