Competition link: https://www.kaggle.com/c/mlb-player-digital-engagement-forecasting
Here is the Code for the project, where:
mlb-explorer.ipynb:
pre modeling researching, focus on
- Targets basic stat or infomation, for example: correlation, distribution, time series ACF/PACF trend,
- Targets distribution with some key features, like player status, have competition on the day.
model-tune-train-model-and-data-export.ipynb
Data clean, selection to training data, feature variable building, modeling stepup. Build the final model which will apply in mlb-trained-model-submission.ipynb.
P.S. non-runned code after model trained can be igorne, which is further study on parameter and features.
mlb-trained-model-submission.ipynb
Code on applying to future application data as a final source in competition.