View on GitHub


Time Series Prediction: A Non Linear Approach with Neural Networks

Download this project as a .zip file Download this project as a tar.gz file


The purpose of this library is using neural networks to replicate classical forecast models from the financial industry structurally, like AR(p), MA(q), ARMA(p, q), ARCH(q) or GARCH(p, q). Currently only supports ARMA(p, q).

Getting started

Install the library and run the ARMA example.

pip install neuralforecast
git clone
cd neuralforecast
python examples/