Extract from column Date class only in R-CodePudding

I have a bizarre database with all sorts of date types, such as:

datfra <- data.frame(ExactDate = c(NA, NA, NA, NA, NA),
                     DateEstimated = c("1974", "22/12/1984","<1982","year?","23/05/1981" ))

> datfra
  ExactDate DateEstimated
1        NA          1974
2        NA    22/12/1984
3        NA         <1982
4        NA          year?
5        NA    23/05/1981

Would I would like to achieve is to transfer correct dates (the ones fitting dd/mm/yyyy) to the column ExactDate and leaving the unfitting ones as NA's such as:

> desired_datfra 
   ExactDate DateEstimated
1       NA          1974
2 22/12/1984    22/12/1984
3       NA         <1982
4       NA          year?
5 23/05/1981    23/05/1981

Any help will be welcomed!!

CodePudding user response：

Using as.Date and format, non-matching dates automatically get discarded.

transform(datfra, ExactDate=format(as.Date(DateEstimated, '%d/%m/%Y'), '%d/%m/%Y'))
#    ExactDate DateEstimated
# 1       <NA>          1974
# 2 22/12/1984    22/12/1984
# 3       <NA>         <1982
# 4       <NA>         year?
# 5 23/05/1981    23/05/1981

CodePudding user response：

Here is an option using dmy from lubridate package.

By using this function only the dates with the correct format will be parsed the other will left NA.

library(dplyr)
library(lubridate)
datfra %>% 
  mutate(ExactDate = dmy(DateEstimated))

   ExactDate DateEstimated
1       <NA>          1974
2 1984-12-22    22/12/1984
3       <NA>         <1982
4       <NA>         year?