deaseasonalize temperature data using means

I have a set of data that includes dates (years from 1995-2018 and months (1-12), where dv(:,1)=year and dv(:,2)=month. It also includes temperature (temp). The dataset has 350,000 data points. I would like to find the monthly mean for each year (e.g., Jan 1995, Jan 1996...Jan 2018) and the mean for each year (e.g., 1995, 1996...2018). To deaseasonalize the data, I need to adjust each monthly mean based on the mean for that year. If the yearly mean is greater than the monthly mean then yrmean-mmean. If the monthly mean is greater than the yearly mean, then mmean-yrmean. This should give me a new 128x1 of deasonalized data. This is where I get stuck. Any help is appreciated! Here's what I have so far:

%find yearly mean

G=findgroups(dv(:,1));

meanyr_temp=splitapply(@mean,temp,G); %yearly mean is 17x1 double, note not all years have data

%find monthly mean for each year

H=findgroups(dv(:,1),dv(:,2));

meanmyr_temp=splitapply(@mean,temp,H); %month/year mean is 128x1 double

Answer 1

[Star Strider]

Without your data, it is difficult to write specific code.

From your description, this may work:

Yrv = [ones(12,1)*2017; ones(12,1)*2018];                           % Create Data
Mov = repmat((1:12)', 2,1);                                         % Create Data
Tv = 10 + 0.1*(1:24)' + sin((1:24)'*pi/6)*15;                       % Create Data

Data = [Yrv Mov Tv];                                                % Matrix Of Years, Months, Temperatures
Times = datetime(Data(:,1), Data(:,2), ones(size(Data,1),1));       % Create ‘datetime’ Array
TData = timetable(Times, Data(:,3));                                % Convert To ‘timetable’

YrMean = retime(TData, 'yearly', 'mean')                            % Yearly Mean
MoMean = retime(TData, 'monthly', 'mean')                           % Monthly Mean

DYrMean = timetable2table(YrMean);                                  % Convert To ‘table’
DMoMean = timetable2table(MoMean);                                  % Convert To ‘table’
Deseasonalize = abs(bsxfun(@minus, DMoMean.Var1, DYrMean.Var1'));   % Absolute Value Of The Difference

The ‘Deseasonalize’ assignment takes the absolute value of the difference between the monthly and yearly means for each year. I am not exactly certain what you mean by ‘deseasonalize’, or what you want for ‘deseasonalized’ data, so this is my best guess. It should get you started.