How to limit a estimate coefficient in a linear regression?

Question

Hello everybody,

I would like to know how to limit a estimate coefficient in a linear regression. I have an estimate coefficient value of 1.7 but it actually doesn´t have sense in real life for what I´m doing, so I would like to limit this coefficient to be <= 1. Is it possible? Thank you!

KennethJ · Answer

I don't know of a way to set boundaries for parameter estimates in a linear regression (or in any other predictive algorithm for that matter), but that might just be because it has never come up for me.

If the parameter estimate does not make sense, however, I think you need to revisit your data and the independent variables that you're using in your model. In most cases, I would see this as an indication that you are missing one or more important input features or that there is another issue with the data such as multicollinearity.

Without a deeper understanding of the exact use-case and the data, it is difficult to provide a more specific recommendation.