This repository contains benchmarks results/logs that are used to evaluate PyAF automatic forecasting tool.

Some benchmarks are official forecasting competitions : M1, M2, M3, M4 , NN3 and NN5

Some are internal benchmarks (used mainly to evaluate new PyAF versions against previous ones) : YahooStocks (all stocks by index are forecast for 7 days).

We will run these benchmarks on a regular basis and update this repository when something changes.

Each log file gives the full log of PyAF execution for forecasting one dataset, contains actual and predicted values as well as some performance measures (MAPE, RMSE, ...) and the total execution time.

Scripts used to run the benchmarks are included with PyAF : https://github.com/antoinecarme/pyaf/tree/master/benches

Last generated benchmark reports : https://github.com/antoinecarme/PyAF_Benchmarks/tree/master/reporting/data

Jupyter notebook to be used to update future reports : https://github.com/antoinecarme/PyAF_Benchmarks/blob/master/reporting/bench-debrief.ipynb