This dataset contains 4 years of electrical consumption, generation, pricing, and weather data for Spain. Consumption and generation data was retrieved from ENTSOE a public portal for Transmission Service Operator (TSO) data. Settlement prices were obtained from the Spanish TSO Red Electric España.
Feeding bossting models with 2015--->2017 years as training data and 2018 as testing data
Test size remains the same, train size increases after each split
In time series modelling, the predictions over time become less and less accurate and hence it is a more realistic approach to re-train the model with actual data as it gets available for further predictions.
We adapt the time series data to supervised learning algorithms with a different approach
- Multivariate forecasting function (i.e. using only the previous time-steps of the electricity price vs. also using other features) using a different number of previous time-steps as the features for the models (3, 10 and 25 previous time-steps for all the used features)
- We choose the adequate time lags for the price hourly prediction
-- We concatenate the following datas :
-- We choose the adequate time lag using autocorrelation analysis :