How can I find and put zero for missinf data in a time series?

Hi all

I have a 30 years of hourly data. Some of the data are missing. It means that I do not have data for example for 1 PM and 2 PM of 12th of JAN. But the missing value are a lot and scattered through the 3 years. How can find them and put zero instead of them in a bigger matrix? I have 5 vectors (years, month, days, hour, values). Some of the hours in some days are missing. We can see them by looking at the data but how can I detect them by MATLAB and increase the size of my matrix by putting zeros instead of them?

I'll appreciate in advance, Andrea

[per isakson]

Do you really want to replace missing data with zero?

Answer 1

[Walter Roberson]

For each time that you do have, calculate an hour number (an integral value) starting from 1 for the first time-point. Then

ExpandedData = accumarray(HourNumber(:), DataValues(:));

The result will be a column vector in which the row number corresponds to the hour number, and the entry for the row corresponds to the data value for that time, and there are 0's for any missing times.

Note: with the call done this way, if there happen to be multiple entries with the same hour number, the entries would be summed. Changing to calculate the mean instead is quite simple -- add the arguments [], @mean to the call.

[Andrea]

Thanks, but still I have question. What is the HourNumber? is it something from 1:24 and repeat again?

[Walter Roberson]

The hour numbers should be consecutive, 25 for the 25th hour, and so on.