MATLAB Answers


How to divide timeseries data into seasonal variation

Asked by Niraj Acharya on 8 Sep 2019
Latest activity Answered by Guillaume
on 9 Sep 2019
This is the first instant of my data.
DateTime Global_active_power Global_reactive_power Voltage Global_intensity Sub_metering_1 Sub_metering_2 Sub_metering_3
1/01/2007 0:00 2.58 0.136 241.97 10.6 0 0 0
I want to divide it into summe and winter data.


What is your definition of summer and winter (when do they start and end)?
June, July and August are summer while December, January and February are winter. I have list of date in this form:
1/1/2007 0.00
1/2/2007 1.00
1/3/2007 1.10

Sign in to comment.

1 Answer

Answer by Guillaume
on 9 Sep 2019

Note that if you're indeed using timeseries, you may be better off using timetables instead. They're slightly easier to use and are more powerful.
The following applies to timetables, I believe it's more or less the same syntax for timeseries.
While you can indeed split the data into winter and summer (and discard the rest):
summer = yourtimetable(ismember(month(yourtimetable.DateTime), 6:8), :); %6 to 8 is June, July, August
winter = yourtimetable(ismember(month(yourtimetable.DateTime), [1, 2, 12]), :); %1, 2, 12 is January, February, December
you may be better off adding a new variable named Season of type categorical:
yourtimetable.Season = discretize(mod(month(yourtimetable.DateTime), 12), 0:3:12, 'categorical', {'winter', 'spring', 'summer', 'autumn'}); %the mod is to bring december as first month
you can then do group calculation by season with e.g. groupsummary.


Sign in to comment.