search

mrosol / Nonlincausality

Python package for Granger causality test with nonlinear forecasting methods.
MIT License
70 stars 13 forks source link

Coefficients #2

Open Fabi9610 opened 2 years ago

Fabi9610 commented 2 years ago

Hi!

Thank you for sharing the code, this is really impressive. Is it possible to extract or look into the assumed coefficients for each lag used within fitting the model?

Best regards

mrosol commented 2 years ago

Thank you very much for your appreciation.

The fitted models are returned by the function in a dictionary. In the case of neural networks, the models are created using the Keras library and in the case of ARIMA statmodels library was used. In both cases, it is possible to obtain models weights.

If you would like to use weights to assess the causality for each lag here is some paper, which you might find interesting: https://link.springer.com/content/pdf/10.1007/s11571-017-9453-1.pdf

Best regards, Maciej Rosoł

Fabi9610 commented 2 years ago

Thank you so much for your answer.

Can you please tell me how it is possible to extract the weights to for each lag in the case of nonlincausality?

Best regards

Fabian