Machine Learning vs Statistical Methods for Time Series Forecasting: Size Matters

This repository contains the code to reproduce the experiments reported in the paper entitled "Machine Learning vs Statistical Methods for Time Series Forecasting: Size Matters", by V. Cerqueira, L.Torgo, and C.Soares.

Running the experiments

To run the experiments, you just need to run the source files in the following order:

1. 1-tsdl-build-db.r
2. 2-main.r
3. 3-result-analysis.r

The default is an online analysis, so the experiments may take a while to run.

Contact

Feel free to get in touch with me about any issue. vitor.cerqueira@fe.up.pt