a) Fred uses LASSO regression to reduce the training error in the many different training models he's created using linear regression. The training error is incredibly small (but not zero) while the validation error is larger than ten thousand. Fred believes that the best way to rectify this is to lower the value of the hyper-parameter lambda.

Is this a correct or incorrect practice?

b) Jonathan has a dataset pertaining to housing prices in North America. The feature "house area" is sometimes in square feet or in square meters. To make up for this inconsistency, Jonathan normalizes the feature.

Is this a correct or incorrect practice