从数据帧列表中选择数据帧

Question

I am trying to select data frames from within a long list of data frames, based on whether certain columns are empty.

Here is a reproducible example, along with the code I have written to try to solve this problem. I am using one random variable as an acceptable substitute for another (here, b for c), so I am not bothered if df1 a ends up in both group 1 and group 2.

d1 <- data.frame(a=rnorm(5), b=1:5, c=rnorm(5))
d2 <- data.frame(a=1:5, b=rnorm(5), c = c(NA, NA, NA, NA, NA))
d3 <- data.frame(a=1:5, b=c(NA, NA, NA, NA, NA), c=c(1:5))

my_test_data <- list(d1, d2, d3)
group_1 <- list()
group_2 <- list()

for (i in 1:length(my_test_data)) {
if(!is.nan(my_test_data[[i]]$b)) {
group_1[i] <- my_test_data[i]
}
else if (!is.nan(my_test_data[[i]]$c)) {
group_2[i] <- my_test_data[i]
}
else NULL
}

I get warning messages saying:

Warning messages: 1: In if (!is.nan(my_test_data[[i]]$b)) { : the condition has length > 1 and only the first element will be used

and group 1 and group 2 are identical to my_test_data

All help greatly appreciated.

Answer 1

Couple of issues going on in your sample code.

The reason you are getting the warning is because your if condition is returning a vector.

is.nan(my_test_data[[3]]$b) 
[1] FALSE FALSE FALSE FALSE FALSE`

The second problem you have is that your sample data doesn't have any NaN values. You have NA values...so is.nan() won't find anything anyway.

The third problem you have is that your if / else if / else isn't formatted right. The else if goes on the same line as the curly brackets } else if () {

A fourth problem is that NULL by itself doesn't do anything. You might as well leave out the else NULL because it doesn't do anything. Or you can change it to do something.

Instead of using !is.na() we can check whether TRUE %in% is.na()

for (i in 1:length(my_test_data)) {
    if(TRUE %in% is.na(my_test_data[[i]]$b)) {
      group_1[i] <- my_test_data[i]
    } else if (TRUE %in% is.na(my_test_data[[i]]$c)) {
      group_2[i] <- my_test_data[i]
    } # else {
    #   NULL
    # }
}

Now, your code will still have some strange lists because you are storing NULL values in the list, but I'm not sure if that's ok for what you're doing.

str(group_1)

List of 3
$ : NULL
$ : NULL
$ :'data.frame':    5 obs. of  3 variables:
    ..$ a: int [1:5] 1 2 3 4 5
    ..$ b: logi [1:5] NA NA NA NA NA
    ..$ c: int [1:5] 1 2 3 4 5