Create column to flag rows within a date period in R

问题

I need to create a "flag" column within my main data frame that flags rows where the date is within a specific time range. That time range comes from a second data frame. I think I'm just stuck on the ifelse (or if) statement because there are NA's in the flag column. Maybe ifelse isn't the way to go. Here's some sample data:

    # main data frame
date <- seq(as.Date("2014-07-21"), as.Date("2014-09-11"), by = "day") 
group <- letters[1:4]                           
datereps <- rep(date, length(group))                  
groupreps <- rep(group, each = length(date))    
value  <- rnorm(length(datereps))
df <- data.frame(Date = datereps, Group = groupreps, Value = value)  

# flag time period data frame
flag <- data.frame(Group = c("b", "d"), 
        start = c("2014-08-01", "2014-08-26"),
        end = c("2014-08-11", "2014-09-01"))

# Merge flag dates into main data frame
df2 <- merge(df, flag, by = "Group", all.x = T)

# Execute ifelse statement on each row
df2$flag <- "something"
df2$flag <- ifelse(df2$Date >= as.Date(df2$start) & df2$Date <= as.Date(df2$end), "flag", "other")

The result is that in rows where a "start" and "end" date are specified, "flag" and "other" are labeled, but where "start" and "end" are NA, I get Na values for df2$flag. This happens even when I initiate df2$flag with "something". I want "other" for all values that are not defined as "flag". Look at rows 50:68.

df2[50:68,]

回答1:

If I was doing this I'd skip the intermediate dataframe (df2) and the merge step and use ifelse with | which means OR.

date <- seq(as.Date("2014-07-21"), as.Date("2014-09-11"), by = "day") 
group <- letters[1:4]                           
datereps <- rep(date, length(group))                  
groupreps <- rep(group, each = length(date))    
value  <- rnorm(length(datereps))
df <- data.frame(DateTime = datereps, Group = groupreps, Value = value)

This applies flag to the criteria you specified:

df$flag <- ifelse(df$DateTime >= as.Date("2014-08-01") & df$DateTime <= "2014-08-11" |
                    df$DateTime >= as.Date("2014-08-26") & df$DateTime <= "2014-09-01", 
                  "flag", "other")

Then you can have a look:

df[df$flag=="flag",]

回答2:

Change your last line to:

for (i in 1:nrow(df2)) {
    if (is.na(df2$start[i])) {
        df2$flag[i] = 'other'
    } else if (df2$Date[i] >= as.Date(df2$start[i]) & df2$Date[i] <= as.Date(df2$end[i])) {
        df2$flag[i] = "flag"
    } else {
        df2$flag[i] = "other"
    }
}

Its ugly but it does the job. This code is not vectorized, so its fine for your situation, but would be slow for larger applications.

来源：https://stackoverflow.com/questions/42516632/create-column-to-flag-rows-within-a-date-period-in-r

标签

if-statement