Calculating the value, excluding nan

I was calculating the average stock return of each year using the following code, and it shows error. This is due to the nan value in the stock return column (column 7). I turned the nan values to zero and it works.

However this will affect the result. I want to calculate only the data without the nan value (if the return is nan, exclude that row from calculation). How can I do that?

for k=1993:2021

stockreturn(k-1992) = mean(data_crsp(y==k,7));

%1st element will correspond to 1993, 2nd - 1994 and so on

end

Answer 1

[Matt J]

subset=1993<=y & y<=2021;

datasub=data_crsp(subset,7);
ysub=y(subset)-1992;

stockreturn=splitapply(@(z)mean(z,'omitnan'),datasub,ysub);

[King To Leung]

Thank you for your reply!

Is this code able to replace all of the codes that i provide? No for loop is needed?

What does subset means?

Also, what does this mean "@(z)mean(z,'omitnan')" ?

A bit more information, I was finding the return of each year, and the date data is in column 2, and the return data in column 7.

[Matt J]

Yes, it is meant to replace the whole code including the loop. DId you try it?

"subset" is a logical idnex into the range of y that you were looping over.

"@(z)mean(z,'omitnan')" is an anonymous function.

Answer 2

[Steven Lord]

Consider storing your data in a timetable array, with the dates associated with your data stored as a datetime array. If you do this, you can use the retime function to calculate the mean of the data, specifying the anonymous function @(x) mean(x, 'omitnan') as the 'Method' in your call.