NaNs and regress lead to error?

Dear MATLAB experts,

I'm conducting multiple regressions with the function:

regress

In some of the regressions, data of the dependent variable is missing. And as I have read in the documentation, 'regress' removes all rows that contain NaNs. I was wondering if the corresponding rows of the independent variable will be removed, as the independent variable is formed by the same number of rows as the dependent variable in each regression. In any case, would this also lead to estimation errors?

Edit: imputing the missing values is not an option

I would really appreciate your help, since this is urgent for a project.

Thank you in advance

[KSSV]

You can fill the NaN's using fillmissing.

[chiefjia]

I already did that for some of them, but there is no good alternative for filling in the remaining missing values, since the existing methods lead to estimation errors.

Answer 1

[Prince Kumar]

Hi Chiefjia,

Yes, if the rows contains NaNs, then the corresponding rows of independent variable will also be removed. This means that your training data would reduce.