R：用日期值替换 NA。 如果两列中都存在日期值，请选择最早的日期

Question

I am working with two data frames and have merged them thanks to the advice on other StackOverflow threads. Now I'm struggling to get the values from one column into the other column IF they meet certain conditions.

The date_first_followup column is almost null but does have some values in it due to historical usage. The Activity.Date.x column always has a date in it but it isn't always the same date value as the date_first_followup column.

What I need to do is to have R replace all of the NA's in the followup column with the dates in the Activity column. Whenever a row has values in both columns take the earliest date.

date_first_followup Activity.Date.X
NA                  NA
NA                  NA
6/30/2018           NA
4/13/2015           3/25/2015
NA                  NA
NA                  NA
NA                  3/21/2018
NA                  4/2/2018
NA                  4/13/2018
NA                  NA
4/13/2015           11/6/2016

I don't know if how the fields are formatted affect the results but they start of as factors and I've tried various formatting options under as.Date() or even just leaving it as factor.

#Trying out a coalesce based solution.
#Use Coalesce to include follow up dates from activities whenever there isn't data already
#present in the followup field.
#
testVlookup_df$Activity.Date.x <- as.Date(testVlookup_df$Activity.Date.x, format = "%/m%/d%/Y")
testVlookup_df$date_first_followup <- as.Date(testVlookup_df$date_first_followup, format = "%m/%d/%Y")
#
testVlookup_df$date_first_followup <- dplyr::coalesce(testVlookup_df$date_first_followup, testVlookup_df$ctivity.Date.x)

Other attempts

#12.10.19 attempting to ONLY pull activity dates in when the column is NULL OR when the column has a value that has a later date.
testVlookup_df$date_first_followup <- ifelse(is.null(testVlookup_df$date_first_followup) == TRUE,
                                             as.Date(testVlookup_df$Activity.Date.x, format = "%m/%d/%Y"),
                                             ifelse(isTRUE(as.Date(testVlookup_df$date_first_followup, format = "%m/%d/%Y") > as.Date(testVlookup_df$Activity.Date.x, format = "%m/%d/%Y")) == TRUE,
                                                    as.Date(testVlookup_df$Activity.Date.x, format = "%/d%/m%/Y"), NA)
)

Another attempt

#12.12.19 58th attempt. Invert the ifelses


testVlookup_df$date_first_followup <-
  ifelse(
    isTRUE(
      as.integer(gsub("-", "", testVlookup_df$date_first_followup)) > as.integer(testVlookup_df$Activity.Date.x, "%Y/%m/%d")
    ) == TRUE,
    as.Date(testVlookup_df$date_first_followup, format = "%m/%d/%Y"),
    ifelse(
      is.null(testVlookup_df$date_first_followup) == TRUE,
      as.Date(testVlookup_df$Activity.Date.x, format = "%m/%d/%Y"),
      NA
    )
  )
class(testVlookup_df$date_first_followup)

Yet another attempt

  #Attempts for if command  using elses. This seems to work but does not loop through the entire vector.                                            
    # testVlookup_df$date_first_followup <- if(is.null(testVlookup_df$date_first_followup)){ as.Date(testVlookup_df$Activity.Date.x, format = "%m/%d/%Y")
    # } else {if(as.integer(testVlookup_df$date_first_followup) > as.integer(testVlookup_df$Activity.Date.x)) {as.Date(testVlookup_df$Activity.Date.x, format = "%m/%d/%Y")
    # } else { NA }
    # }
    # Got this message 'Warning message:In if (as.integer(testVlookup_df$date_first_followup) > as.integer(testVlookup_df$Activity.Date.x)) 
    # { : the condition has length > 1 and only the first element will be used' Looked it up. Tells me if and else can only be used for 1 argument will not loop. Need sapply or ifelse

Answer 1

When asking a question you should provide a sample of the data, especially because from your question it is hard to tell whether the problem is data formatting, or execution of the task. If the data format is not a problem and R is recognizing the variables as dates (which you can check by passing the column to the str() function) then this should work.

#Make example data
sampleData <- data.frame(
  date_first_followup = sample(seq(as.Date('1999/01/01'), as.Date('2000/01/01'), by="day"), 12),
  Activity.Date.x = sample(seq(as.Date('1999/01/01'), as.Date('2000/01/01'), by="day"), 12)
)

naNums <- sample.int(12, 5)
sampleData[naNums, 1] <- NA

# Replace values
for (i in 1:length(sampleData$date_first_followup)){
  ifelse(is.na(sampleData[i, 1]),
     sampleData[i, 1] <- sampleData[i, 2],
     sampleData[i, 1] <- pmin(sampleData[i, 1], sampleData[i, 2]))
}

Answer 2

I finished this project a few months ago. So here is what eneded up working.

# STORE DATES AS DATES
#Store activity and followup dates as actual dates 
activities_uniq_df$Activity.Date <-as.Date(activities_uniq_df$Activity.Date, format = "%m/%d/%Y")
class(activities_uniq_df$Activity.Date)

main_df2$date_first_followup <-as.Date(main_df2$date_first_followup, format = "%m/%d/%Y")
class(main_df2$date_first_followup)


#Create a subset of the main activities_uniq dataframe that only contains "Follow Up" activities. 

FollowUpSub <- activities_uniq_df[activities_uniq_df$Item == "Follow Up",]

### PULL ACTIVITY DATES from Activities dataset into Main dataset based on Case ID#.

#FOLLOW UPS


#1st pull all of the desired fields in
main_df2 <- merge(main_df2, FollowUpSub[ , c("Item", "Activity.Date", "Case.ID")],
                        by.x = "Matter.Case.ID.", by.y = "Case.ID", all.x = TRUE)

which( colnames(main_df2) == "Activity.Date")

#2nd Re-Format the activity and followup dates because the merge causes date to go back to factor
main_df2$Activity.Date.x <-as.Date(main_df2$Activity.Date, format = "%m/%d/%Y")
class(main_df2$Activity.Date)

main_df2$date_first_followup <-as.Date(main_df2$date_first_followup, format = "%m/%d/%Y")
class(main_df2$date_first_followup)

#class(main_df2$date_first_followup) <- "Date"


#Finally do the actual replacing process
main_df2$date_first_followup <- ifelse(is.na(main_df2$date_first_followup), main_df2$Activity.Date.x, 
                                             ifelse(main_df2$date_first_followup > main_df2$Activity.Date.x, 
                                                    main_df2$Activity.Date.x, main_df2$date_first_followup )
)

#The above step messes with followup date format again so may have to re-class it. 
#Question does the format of dates affect what date is considered less than or greater than another?
class(main_df2$date_first_followup) <- "Date"

#Confirm successful replacements
main_df2[c(35:39, 7001), c(26, 104)]
names(main_df2[c(22, 104)])