search

nredell / forecastML

An R package with Python support for multi-step-ahead forecasting with machine learning and deep learning algorithms
Other
130 stars 23 forks source link

Adding other performance/error metrics and saving model output #47

Open adeel1997 opened 3 years ago

adeel1997 commented 3 years ago

Great package Nick! I am wondering how I can find other performance metrics like R2, bias. Another thing I noticed when I was saving the model output is that the size of the model is quite big. How I can save the model data in a compressed format so that I can use it again?

adeel1997 commented 3 years ago

Okay, I figured how to add error metrics based on aggregate function. One has to group by model,forecast_horizon and window then calculate the defined error metric(R2, Bias). Finally just take a median(aggregate function) of the error metric.

adeel1997 commented 3 years ago

Would be good to know your thought on how to save the model results? So that it can be reproduced in the next session.

Finally, if someone has more code snippet of applying caret based machine learning model in the functional wrapper used in this package. It would be good to look at it.