machine-learninglogistic-regressionevaluation

Does an accuracy of 97.55 % indicate an overfitted model?

I have implemented a logistic regression model in predicting high-risk claims. My model is giving an accuracy of 97%. Does that indicate that my model is overfitted?

Solution

  • The accuracy in the training set is not enough for the overfitting. Are you using a validation set? If you are not using it, you should split your data set to three parts: Validation set, test set and training set. During the training, you should use training and validation set. If you have high accuarcy in training set and low accuracy in the validation set then your model is overfitting.