R. 如何创建一个新列，根据 R 中的另一列返回 i

Question

I have a dataframe

employee <- c('John Doe','Peter Gynn','Jolie Hope')
salary <- c(21000, NA, 26800)
startdate <- as.Date(c('2010-11-1', NA,'2007-3-14'))

employ.data <- data.frame(employee, salary, startdate)

I want a new column employ.data$NA that returns i in employ.data$employee if [i] in any other row is NA.

I have tried this for 1 column but getting errors

employ.data$NA = NA 
{for (i in 1:nrow(Eurostat)) 
  {
  if (startdate[i] = "NA")  employ.data$employee[i]
}

Any help would be appreciated.

Answer 1

You need complete.cases() from base R -

employ.data$missingFlag <- !complete.cases(employ.data)

    employee salary  startdate missingFlag
1   John Doe  21000 2010-11-01       FALSE
2 Peter Gynn     NA       <NA>        TRUE
3 Jolie Hope  26800 2007-03-14       FALSE

Answer 2

Try to vectorize it and use an ifelse statement:

employ.data <- data.frame(employee, salary, startdate, stringsAsFactors = F)
employ.data["missing"] = with(employ.data, ifelse(is.na(startdate), employee, NA))
employ.data
    employee salary  startdate    missing
1   John Doe  21000 2010-11-01       <NA>
2 Peter Gynn     NA       <NA> Peter Gynn
3 Jolie Hope  26800 2007-03-14       <NA>

Alternatively, to check all columns, use any :

employ.data <- data.frame(employee, salary, startdate, stringsAsFactors = F)
employ.data["something_missing"] = apply(employ.data, 1, function(x) any(is.na(x)))
employ.data
    employee salary  startdate something_missing
1   John Doe  21000 2010-11-01             FALSE
2 Peter Gynn     NA       <NA>              TRUE
3 Jolie Hope  26800 2007-03-14             FALSE

The construct above will give you booleans. If you want to get a column of the names, you can combine it with ifelse .

On a more general note, instantiating a column first and then looping through the dataframe to populate it is not particularly Rtistic , and I would suggest to avoid this strategy whenever possible. The apply -family of functions are very powerful, and ifelse is too. dplyr 's mutate combined with case_when statments can also be used in case you want something more SQL-like.

Just for pedagocial reasons, here is your code in working version. Please don't use it, just try to understand the differences.

employ.data$missing = NA 
for (i in 1:nrow(employ.data)) {
  if (is.na(employ.data$startdate[i])){ 
         employ.data$missing[i] <- employ.data$employee[i]
        }
}

Importantly, note that "NA" is interpreted as a string. To test if a value is NA , you need to use eg is.na . After all, testing if 42 == NA is ambiguous. The value is missing. It may or may not be equal to 42, so the test will return NA .

Answer 3

It can be done quite easily with dplyr:

library(dplyr)

employee <- c('John Doe','Peter Gynn','Jolie Hope')
salary <- c(21000, NA, 26800)
startdate <- as.Date(c('2010-11-1', NA,'2007-3-14'))

employ.data <- data.frame(employee, salary, startdate)

employ.data <- employ.data %>% 
  rowwise() %>% 
  mutate(missing = any(is.na(c(salary, startdate))))