Can a discrete variable be a dependent variable in a linear regression problem? If so, how could we control the predictor variable to be a bounded one

You can create a one-hot encoded vector out of these discrete variables.

Hi, sorry !! Could you please be a bit more descriptive? So you mean to say that my predictor variable has to go through one hot encoder so that it will not unbounded values?

So sorry, i misunderstood your doubt to a classification problem. I was trying to say if you have limited number of such discrete outputs, you can create some encodings for them.

If you mean to say whether we can get an output as a discrete value by using linear regression, the answer is yes we can.

Thanks much Ishvinder