I am trying to apply ridge regression to a telecom churning data where dependent variable, Churn is a categorical variable, there are only 3 continuous predictor variables out of 30, where Multicollinearity exists somewhat. Can I use ridge regression?

CodePudding user response：

It seems that you are looking for a multinominal ridge regression:

The short answer is Yes!

See here for further explanation: https://stats.stackexchange.com/questions/104779/why-does-ridge-regression-work-well-in-the-presence-of-multicollinearity

And consider the advice by Peter Flom regarding excact collinearity.

See also here: Multinomial ridge regression using glmnet gives results with probabilities above one and negative for different target classes, any idea why? for an example.

CodePudding user response：

As your task is to classify the customers as Churn or Non-Churn, so its a classification task and not a regression task.

Ridge regression is used for regression purpose only as it needs the dependent variable to be continuous.

So for your analysis Ridge regression can't be used.

Special characteristic of Ridge regression is it works fine in presence of multicollinearity but with a continuous dependent variable.

You may refer this https://en.wikipedia.org/wiki/Ridge_regression