CODEWITHSUNDEEP
rdataframe
Peteris
Peteris

Reputation: 3756

Trying to map timestamps to dates in R data frame

I have a data frame df with a column time that contains timestamps and a function from.timestamp

from.timestamp = function(timestamp) {
  return(as.POSIXlt(timestamp/1e3, origin="1970-01-01 00:00:00 GMT"))
}

that converts a timestamp to a date type. How do I modify the data frame so that all timestamps are converted to dates?

My approach

I'm trying the following:

df$time <- apply(as.array(df$time), MARGIN=1, FUN=from.timestamp)

The resulting column entries for time do not look like a date, but are rather a sequence of numbers, e.g., 1.034, 59.000, 23.000, 7.000, 5.000, 113.000, 5.000, 157.000, 1.000.

dput(df) gives:

structure(list(time = c(1370674741034, 1372671995085, 1370847555008, 1371456058556, 1372570911379, 1371937835807)), .Names = "time", row.names = 8831395:8831400, class = "data.frame")

Upvotes: 1

Views: 929

Answers (1)

Hong Ooi
Hong Ooi

Reputation: 57686

Just do

df$time <- from.timestamp(df$time)

The reason you're getting the strange behaviour is because POSIXlt objects are actually lists, with components for year, month, day, etc. In your code, apply receives from from.timestamp a sequence of lists, which it combines into a single list. In doing so, it strips the class information from the result, which is why you saw your odd numbers. These odd numbers are actually the individual components of the POSIXlt object representing that datetime: seconds, minutes, hour, day of month, month, year, weekday, day in year.

Upvotes: 2

Related Questions