search

bukosabino / btctrading

Time Series Forecast with Bitcoin value, to detect upward/down trends with Machine Learning Algorithms
MIT License
154 stars 47 forks source link

Walk Forward Model Validation #8

Open slavakurilyak opened 7 years ago

slavakurilyak commented 7 years ago

I would love to see the implementation of walk-forward testing.

A rolling-forecast scenario also called walk-forward model validation is the recommended method (gold standard) for constructing machine learning models dedicated to time-series data. Walk-forward validation provides the most realistic evaluation of machine learning models since k-fold cross validation does not work for time series data.

Each time step of the test dataset is walked one at a time. A model is used to make a forecast for the time step, then the actual expected value from the test set is taken and made available to the model for the forecast on the next time step.

For more info, see code examples (1, 2) or research paper.

bukosabino commented 7 years ago

Hi @slavakurilyak ,

This is a excellent idea to develop on this project! But, I need more time.. :P

Thank you