Converting selected columns from POSIXct to Date with particular format using apply function

user_9401203 0
Viewed 1487

Occasionally, I read in data from Microsoft Excel into R. In Excel, the date variables are formatted correctly (e.g. 31-Dec-2017). Following reading into R, the same date variable gets converted into another format (e.g. 2017-12-31).

A sample of my data frame (after being read into R) looks like the following:

df <- structure(list(ID = c("001", "002", "003", "004", "005"), 
           t1_date = structure(c(1490227200, 1508198400, 1511395200, 1527292800, 1485216000), 
                                         class = c("POSIXct", "POSIXt"), tzone = "UTC"), 
           t2_date = structure(c(1524009600, NA, NA, NA, 1523232000), 
                                         class = c("POSIXct", "POSIXt"), tzone = "UTC"), 
           t3_date = c(NA, NA, NA, NA, NA), 
           t4_date = c(NA, NA, NA, NA, NA), 
           t5_date = c(NA, NA, NA, NA, NA)), 
      .Names = c("ID", "t1_date", "t2_date", "t3_date", "t4_date", "t5_date"), 
      row.names = c(NA, -5L), class = c("tbl_df", "tbl", "data.frame"))

Using the t1_date variable as an example, I can individually convert each date column from the POSIXct class (which R selects for me) into the Date class and then changing the format using the following code:

df$t1_date <- as.Date(df$t1_date)
df$t1_date <- format(df$t1_date, "%d-%b-%Y")

However, I have many date columns and they may not be adjacent to one another in the data frame.

Normally, to solve such problems, I use the which and apply functions to select the relevant columns and apply the function to them:

df[, which(colnames(df) == "t1_date"):
 which(colnames(df) == "t5_date")] <- 
  apply(df[, which(colnames(df) == "t1_date"):
         which(colnames(df) == "t5_date")], 2, function(x) as.Date(x, format = "%d-%b-%Y"))

The above code resulted in NAs in all the date columns and I am unsure why. Even if I do not force the change in format, but convert the date columns from the POSIXct class to the Date class, the code still does not work (see below):

df[, which(colnames(df) == "t1_date"):
 which(colnames(df) == "t5_date")] <- 
  apply(df[, which(colnames(df) == "t1_date"):
         which(colnames(df) == "t5_date")], 2, as.Date)

My desired result is to convert the dates to a DD-MMM-YYYY format (e.g. 31-Dec-2017).

Help appreciated. Thanks!

posixct apply r date
1 Answers

We select the range of columns which we want to format and then convert those into dates by using as.Date and then format it based on our requirement.

start_col <- which(colnames(df) == "t1_date")
end_col <-  which(colnames(df) == "t5_date")
df[start_col:end_col] <- lapply(df[start_col:end_col], 
                        function(x) format(as.Date(x), "%d-%b-%Y"))


df
# A tibble: 5 x 6
#  ID    t1_date     t2_date     t3_date t4_date t5_date
#  <chr> <chr>       <chr>       <chr>   <chr>   <chr>  
#1 001   23-Mar-2017 18-Apr-2018 NA      NA      NA     
#2 002   17-Oct-2017 NA          NA      NA      NA     
#3 003   23-Nov-2017 NA          NA      NA      NA     
#4 004   26-May-2018 NA          NA      NA      NA     
#5 005   24-Jan-2017 09-Apr-2018 NA      NA      NA

The same can also be achieved with dplyr , mutate_at

library(dplyr)
df %>%
  mutate_at(vars("t1_date":"t5_date"), funs(format(as.Date(.), "%d-%b-%Y")))


#  ID    t1_date     t2_date     t3_date t4_date t5_date
#  <chr> <chr>       <chr>       <chr>   <chr>   <chr>  
#1 001   23-Mar-2017 18-Apr-2018 NA      NA      NA     
#2 002   17-Oct-2017 NA          NA      NA      NA     
#3 003   23-Nov-2017 NA          NA      NA      NA     
#4 004   26-May-2018 NA          NA      NA      NA     
#5 005   24-Jan-2017 09-Apr-2018 NA      NA      NA
user_39629140
Related