All code in the MSBD5001-Fall2019 In class competition for MSBD 5001.ipynb file.
And the result submition is sub_PassiveAggressiveRegressor.csv.
I build a pipeline to try 10 Regressor model from sklearn and Xgboost, finally chioced the PassiveAggressiveRegressor model.
And luckily get the rank 2/86 on the public leadborad.
Actually this data is not very suitable for machine learning. Because we use the video games information to predict how long time a specfic person's time on it. It's too random and the correlation between X and y is too small I think.
What makes me get the rank 2 position is really depends on lucky!