when I am doing training data I get accuracy above 70 but when I test it against the test case it achieve only aprox 68 percent and when i get 56 percent accuracy in training data then i am getting above 72 percent accuracy on test case what is going on??
Diabetes Classification Challange
Hi @Nikhil799
The first problem you reported is a case of over-fitting. Simply put, over-fitting is when your model just learns the training data without recognising any patterns and hence, performs poorly on unseen test data.
The second generally occurs due to noise, random state, less testing data etc. It’s hard to pin point just from the accuracy numbers. Try training again and let me know if the problem persists!