I am working on classification problem where I have categorical and continuous features however the target is binary.

  1. What is the best way to check correlation with respect to target variable.

  2. Also I want to eliminate some features which are highly correlated.

I have generated a correlation heatmap for continuous features and for categorical features I am thinking to perform a chi square test .

Is this a right approach? Do I have to apply two different methods for continuous and categorical features?


For correlations between continuous and categorical variables see Correlations between continuous and categorical (nominal) variables and Correlations with unordered categorical variables.

But your main question seems to be about classification into two classes, since the target is binary. It is in most cases better to see this as a risk estimation problem, and that can be modeled via logistic regression, see Choosing between logistic and discriminant, Why isn't Logistic Regression called Logistic Classification?, When is logistic regression suitable?

And finally, you want to eliminate features. Doing that by bivariate correlation/association is seldom a good idea. See How to reduce predictors the right way for a logistic regression model.

| cite | improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy