How is calculated the classification rate?
We have made a XGB model and obtained a accuracy = 0.88 on validation data ( 30% from the original training set) but when we have sent the result we have obtained only 0.71. why?
Maybe you are overfitting?
What are your exact parameter settings?
@hammerproject I shared my XGBoost solution in Share Your Approach thread. It scored .8247. Feel free to ask me any questions about my methodology.
here you can find our modelhttps://github.com/HammerProject/PumpIt-DrivenData/
at the begin the model overfitting --> we don't use cross-validation and use all the data for training
the final model scored 0.7688