Feature engineering


I wanted to do some feature engineering and add them as my features for my models. I know I can incorporate this process in doing my local training and testing but I am trying to wrap my head how this (I meant the data processing to create the new features) can be incorporated to the hidden validation data set. In short, can we do feature engineering for this competition? If yes, is there a way to incorporate the logic to the hidden validation data set? Thank you!

I think we can do feature engineering within the train() and infer() method. These methods are called when our code is executed on the competition server to calculate scores on unknown data.

1 Like