Classification Hackthon using AUC_ROC evaluation



Hello AV’s
i’m new to machine learning technology. recently i participated my first hackthon but i was stuck middle of the program.can any one guide me on this. as per my understanding to find roc_auc score we use(y_test,probs[1]) but here we doesn’t have y_test values, in that scenario how do we check roc_auc score. for more info on this prob pls chec below URL.



Hi @Arjun_1987,

In order to check the auc score on your system, you can create a validation set using train_test split. For getting the auc on your actual test set, submit your predicted values and you will get your score.


Hello there @Arjun_1987,

The y_test is actually the test sample part of the trainset of X and y and not the actual test set . Also, I see that your are indexing the probs with [1] using probs[1] which justs indexes single value. Instead you should index something like this --> probs[;,1]