Say I have the following dataset:
| A | B | C |
|---|---|---|
| 1 | NA | NA |
| NA | 2 | NA |
| NA | NA | 3 |
and I am trying to get
| A | B | C |
|---|---|---|
| 1 | 2 | 3 |
The context for this is that I have such a dataset, where there are values if A is non-empty, B/C are always NA and same goes for B/C. Moreover, the lengths of ABC are not the same.
These are values I then use to calculate overlap coefficient with overlap(x,y) from library(bayestestR). However, if I leave NAs in, the results for overlap(a,a), i.e. the same set, are not 1 but .98 or something along those lines. I am not sure what to do. Please advise! Thanks in advance.
If someone has a tip on how to prompt overlap() to ignore NAs, that would work as well I guess!
I tried
df <- data.table(df)[, lapply(.SD, function(x) x[order(is.na(x))])]
df[!result[, Reduce(`&`, lapply(.SD, is.na))]]
df[complete.cases(df), ]
But it still left NA values, even plenty of rows where all values are NA! This bit of code helps to move up all the non-NA values to the top, but NAs are still there.
If your data only have one numeric value in each column, you can use colSums with na.rm = TRUE. If you want the output in a data.frame, wrap it in data.frame(as.list(...)):
data.frame(as.list(colSums(df, na.rm = TRUE)))
# A B C
# 1 1 2 3