How to change 3:00 to minutes:seconds date time in R?-CodePudding

I have a dataset in R that contains minutes and seconds in the following format:

3:00

As of now it is a character, how can I change it to a date time so that it can be analyzed in a graph? I want to analyze posts on instagram. I have their likes and video time and want to plot them. I cant do this unless I believe I turn it into a date time?

CodePudding user response：

Not clear what is meant by date time when we only have minutes and seconds but these two turn x into a times object, x1, and a lubridate Period object, x2, respectively. Internally a times object is a fraction of a day and as.numeric(x1) will give that fraction. as.numeric(x2, unit = "sec") will give the number of seconds corresponding to the Period object.

library(chron)
x <- "3:00"
x1 <- times(paste0("0:", x))

library(lubridate)
x2 <- ms(x)

CodePudding user response：

I imagine that your data frame looks something like this example I created.

df <- data.frame(time= c("3:00", "4:00", "2:00", "1:00", "6:00", "3:30", "1:10"), video= c(1:7),
                 likes= c(10, 17, 6, 25, 14, 5, 19))

library(lubridate)
library(tidyverse)

I suggest using the lubridate package to convert to seconds.

df2 <- df %>%
  mutate(time_sec= as.period(ms(time), unit = "sec"))

Then you might consider presenting it as a scatter plot with likes versus time.

ggplot(df2, aes(x= likes, y= time_sec))  geom_point(size=2, shape=23)

CodePudding user response：

You could convert time to decimal.

t2dec <- \(x, type=c('mins', 'secs')) {
  type <- match.arg(type)
  r <- x |> 
    strsplit(':') |>
    do.call(what=rbind) |>
    type.convert(as.is=TRUE)
  if (type == 'mins') {
    r[, 1]   r[, 2]/60
  } else {
    r[, 1]*60   r[, 2]
  }
}

t2dec(dat$time)  ## as minutes (default)
# [1]  1.583333  6.066667  7.366667  8.283333  8.850000 15.616667 23.383333 28.033333 33.350000 36.150000
# [11] 37.550000 40.250000 41.933333 42.700000 43.550000 47.833333 49.283333 55.066667 56.600000 59.883333

t2dec(dat$time, 's')  ## as seconds (since "s" matches "seconds")
# [1]   95  364  442  497  531  937 1403 1682 2001 2169 2253 2415 2516 2562 2613 2870 2957 3304 3396 3593

Accordingly:

op <- par(mfrow=c(1, 2))
plot(t2dec(dat$time), dat$x, type='b', col=4, xlab='mins', ylab='likes', main='Videos')
abline(h=0)
plot(t2dec(dat$time, 's'), dat$x, type='b', col=4, xlab='secs', ylab='likes', main='Videos')
abline(h=0)
par(op)

Note: R >= 4.1 used.

Data:

dat <- structure(list(time = c("01:35", "06:04", "07:22", "08:17", "08:51", 
"15:37", "23:23", "28:02", "33:21", "36:09", "37:33", "40:15", 
"41:56", "42:42", "43:33", "47:50", "49:17", "55:04", "56:36", 
"59:53"), x = c(-2L, 2L, -3L, 3L, 7L, 7L, -3L, 0L, 1L, 5L, 3L, 
0L, 1L, 10L, 1L, 1L, 6L, 3L, -1L, -1L)), class = "data.frame", row.names = c(NA, 
-20L))