OK to use residual sum of squares for cross-validation of binary outcome?

Question

For an OLS model the mean squared error can be used to assess the fit of the trained model on the validation data.

What is the equivalent for a logistic regression model? Can I simply use the following residual sum of squares function?

$RSS=\Sigma^N_{i=1}(y_i-{\hat p})^2$

Answer 1

What you propose is equivalent to using the Brier score which is a proper scoring rule. So it is certainly OK to use it for cross-validation.

See also logistic regression predictive modeling.