Now that the contest is over please reveal your approach for Loan Prediction III. It seems the public leader board score is stuck with some discrete values. I’ve go score of 0.791667. I performed some feature engineering like log transformation of big values instead of scaling, creating features like monthly loan amount from
LoanAmount. And some other feature engineering too. I used an ensemble model of XGboost, Random Forest and Adaboost. I got my maximum cross validation score with
StatifiedKFold of ~ 89% and without
KFold of ~ 83%. I shall upload the code soon for all the beginners. Please share your approach and code for all who got better score. Also do you guys think it was more of a feature engineering problem?