Reshape hourly mean values with missing data/ non-uniform data

Hi, I could get the hourly averages from minute interval data, using unique and accumarray functions.

I would like to arrange the averaged values in a matrix but there might be some data (hours) missing.

For e.g. my output matrix should look like:

yy mm dd hr1 hr2 hr3 ... 0 0 0 hr20 ... 0
yy mm dd 0   0   0   ... hr10 ..... 0 0 hr24

Is there a way to identify the missing hours data and put zeros there and to create this matrix...!

Thank you.

[Walter Roberson]

Do you have a new enough version of MATLAB that you could switch to timetable() objects instead? R2016b or newer I think it is.

[Venkata]

No, I am using R2016a. Is there any work around?

Answer 1

[Walter Roberson]

Supposing that T is a vector of datenum for each reading (the timestamp) and D is the corresponding value, then

dv = datevec(T);
days = datenum([dv(:,1), dv(:,2), dv(:,3)]);
[unique_days, uidxa, uidxc] = unique(days);
hours = dv(:,4);

summary_data = accumarray( [uidxc, hours+1], D(:), [], @mean);

summary_table = [dv(uidxa,1:3), summary_data];

[Venkata]

Thanks a ton, Walter Roberson.

It worked like a charm.