R: Splitting a Time Series into custom seasons

Question

I have a time Series DataFrame:

[https://www.dropbox.com/s/elaxfuvqyip1eq8/SampleDF.csv?dl=0][1]

My intention is to divide this DataFrame into different seasons according to:

1. winter: Dec Jan Feb
2. Pre-monsoon: Mar Apr May Jun15 (i.e. till 15th of June)
3. Monsoon: 15Jun Jul Aug Sep (i.e. from 15th of June)
4. Post-monsoon: Oct Nov.

I tried using openair package function

selectByDate()

But no luck yet. Being novice in R. Any help would be highly appreciated.

Thanks!

Answer 1

Please see the lubridate package which makes working with date/time a bit easier.

For your problem, I guess you can use sapply:

df["season"] = sapply(df["date"], assign_season)

where, assign_season:

assign_season <- function(date){
    # return a season based on date 

}

once you have seasons, then you can divide the dataframe easily:

winter = subset(df, season == "winter")
# and so on

Sorry, I have to rush now, but can come back and finish this, if someone else hasn't answered already.

EDIT:

So, R does have a built in function cut, that can work on dates and split a vector based on date ranges.

For your data, I did this like so:

library(lubridate)
library(dplyr)

df = read.csv('SampleDF.csv')

## reformat date into POSIXct
df <- df %>%
         mutate(date_reformat = as.POSIXct(date(mdy_hm(date))))

## define breaks & labels
breaks = c("2014-12-01", "2015-03-01", "2015-06-15", "2015-10-01", "2015-12-01", "2016-03-01", "2016-06-15", "2016-10-01", "2016-12-01", "2017-03-01")
labels = c("winter", "pre_monsoon", "monsoon", "post_monsoon", "winter", "pre_monsoon", "monsoon", "post_monsoon", "winter")
df["season"] = cut(df$date_reformat, breaks=as.POSIXct(breaks), labels=labels)

splits = list()

for (s in c("winter", "pre_monsoon", "monsoon", "post_monsoon")){
  splits[[s]] = subset(df, season == s)[c("date", "value")]
}

Now, the splits list should have all the data you need