Question on Machine Learning(Linear Regression)

What if ,we remove all the feature columns from the training data which has the negative impact on RMSE?
Does our model will get better or worse?

If you talk about the coef_ values, then I would advise not to remove such columns.

It’s important that the model can learn that some values have a negative impact on the price.

Example:
Having a nice, big house, with a pool, and all the stuff you ever dreamed of in proximity of radioactive waste dump wouldn’t be so nice, so you can expect the price to drop in such case considerably. If you never considered the negative impact, you would see a price that doesn’t exactly reflect the value of such estate.

1 Like