How to use splitapply with fitlm

Hi,

I want to run a regression for 2 diffrent stocks for each year of observation separately. So i get a new Betas for each stock and each year. The goal is to assign the residuals to ID and dates. I upload the set I'm using. The regression looks like this:

RET = Beta0 + Beta1 * MRET

RET is the returns of the stocks and MRET the return of the Market.

My idea was to you use the splitapply function. But my code isn't working. I guess the problem is in line 3.

[group6, years, ID]= findgroups(subset1.years, subset1.ID);
[group7, dates]= findgroups(subset1.years, subset1.ID);
regression = @(y) fitlm(subset1.MRET,subset1.RET);
lm2 = splitapply(regression, subset1.MRET,group6);
Resitbl = table (ID,dates, lm2.Residuals.Raw); 

regress function would also work for me. But it also didn't work for me.

Thanks in advance

Answer 1

[Matt J]

Your regression(y) function takes an argument y that is never used.

[Luca]

Hi Matt,

Thanks for your answer.

regression = @(subset1.RET) fitlm(subset1.MRET,subset1.RET);

Is this the right way to correct it ?

[Matt J]

This might be what you want,

lm2 = splitapply(@(x,y){fitlm(x,y)}, subset1.MRET, subset1.RET,group6);

[Luca]

Thank you very much the code is working now.

But there is a new problem.

I get a tabe for lm 2 where each regression is stored in one row. I attached a foto.

And I'm not able to access the residuals all at ones. Do you know a way?

Regards

[Matt J]

In the photo you've shown, lm2 is a cell array, not a table.

[Luca]

Right, sorry I did not pay attention.