This work was done as part of a practical work for the Machine Learning class at Ecole des Mines de Saint-Etienne in France.
- TP_2_TimeSeries_preprocess.py : data-preprocessing script, not modified.
- TP_2_TimeSeries-tbc.py : SVR & ANN for reconstruction task.
- TP2_timeSeries_prédiction.py : ANN for prediction task.
The objective of this practical work was to propose machine learning models on two datasets : resultsSolar.csv and resultsWind.csv which are energy production data from a fictive alternative energy provider. For each data set, two tasks were done : reconstruction (prediction at a current or past time) and prediction (prediction for future time).
Both of these tasks were done on two different features: 'System power generated (kW)' and 'Electricity load (year 1) (kW)'.
A complete report is available (in French) here