- README. md This file
- requirements.txt The packages you need to install at first.
- forecasting.ipynb
The main program built in jupyter-notebook.
- Link for nbviewer : forecasting.ipynb
- app. py The main program.
- next_week.csv The estimated results of power supply for the next week from Taiwan Power Company.
- power_dataset.csv The power usage dateset ( 2017/01/01 ~ 2019/02/28 ) from Taiwan Power Company.
- submission.csv The predicted peak load for the next week. ( 2019/04/02 ~ 2019/04/08 )
Use svm-svc model from scikit-learn. The training data is '淨尖峰供電能力' and '尖峰負載' from the power_data set. The reason of choosing these feature is that '淨尖峰供電能力' is the estimated value of maximum of '尖峰負載' from Taiwan Power Company. So it is the most related feature to the '尖峰負載', which we want to predict.