Feature Scaling for Single input variable

Do we need to apply feature scaling if we are fitting a linear regression model on a data that only has single input variable or single feature with a range of 200 - 30,000?

Besides converging faster, if we do scaling, we can use our rule-of-thumb learning rate to do most of the training for neural network of similar architecture.

I think these are some benefits of scaling. If you agree on the above, do you think we “need to”?



So we need to perform feature scaling for univariate linear regression?

Hello @realarslan33,

We can choose to, in order to have the benefits.

1 Like