Regularization-regression

Situations where regularization is useful

• Overfitting

• Curse of dimensionality

• High covariance between variables

• Underfitting

The two most common types of regularization are

• Lasso

• least square

• Σ

• Ridge

The Lasso is particularly useful when you have:

• redundant variables

• unimportant variables

• Underfitting

The Ridge penalty will

• adds the sum of the squared (non-intercept!) values to the loss function

• subtract the sum of the squared (non-intercept!) values to the loss function

• remove the sum of the squared (non-intercept!) values from the loss function

• adds the sum of the absolute values of the coefficients

Elastic Net is simply a combination of the Lasso and the Ridge regularizations.

• False

• True

The term multicollinearity means

• there are low correlations between predictor variables in the model.

• there are high correlations between predictor variables in your model.

With the Lasso and Ridge it is necessary to ______ the predictor columns before constructing the models

• Centralize

• standardize

The Ridge is faster to compute

• True

• False