The data is about loan performance (default or not) - it has 9000+ rows and very clean. You need to build a classifier (programmed in Python) with Random Forest and XGboost and do the usual evaluation based on CV. The features will be given to you, no feature engineering/selection on your end. The only two slightly fancy requests - 1. when building the model, down sample the non-default group as the data is high unbalanced. 2. do some grid search of hyper-parameters for the Random Forest and XGboost models ( just two parameters for each model will do). This should take no more than 5 hours for an experienced data scientist.
It is very urgent and needs to be done in the next 10 hours. Thank you!