30-day running mean from hourly data

Hello, I have a hourly meteorological data as attached. The first column is year, 2nd is month, 3rd is day, 4th is hour and the last is the value. I need to calculate the 30-day running mean from the hourly data and then subtract hourly data from the running mean values. Any help?

Answer 1

[KL]

dt = datetime([Yrly_slr(:,1:4) zeros(length(Yrly_slr),2)]);
TT = timetable(dt,Yrly_slr(:,5));
TT2 = retime(TT,'monthly','mean')

This is not a 30 day running mean but rather the monthly mean. I hope this is what you intended to do

[Poulomi Ganguli]

Hello, I don't want monthly mean. I need to calculate 30-day running mean and subtract them from hourly data. The final output will be time series anomaly = hrly data - 30-d running mean(hrly data);

[KL]

load('slr.mat')
newMat = reshape(Yrly_slr(:,5),30,floor(length(Yrly_slr)/30));
meanMat = repmat(mean(newMat,2),1,floor(length(Yrly_slr)/30));
res = newMat-meanMat;

[Poulomi Ganguli]

First of all, the value in column 5 are hourly values, so there are 24 hours in each day, if we reshape it to 30, it won't convert into 30-d value. I don't think the suggested code is correct. I will wait for some more suggestions. Thanks

[KL]

Yes, that's right but still the idea is the same with reshape!

load('slr.mat')
newMat = reshape(Yrly_slr(1:24*30*12,5),24*30,12);
meanMat = repmat(mean(newMat),24*30,1);
res = newMat-meanMat;

Answer 2

[Steven Lord]

You might find the "Sample Points for Moving Average" example in the documentation for the movmean function to be of interest.

Answer 3

[Nalini KS]

How to do this for data having NaN values