As every time we use Exogenous features we need something to explain a posteriori why a feature brings more value than an other one to the final prediction.
Feature importance or SHAP are some tools that we have in MLForecasts
I would love to have somehing similar in Statsforecast, specially for models that accepts exogenous features (like mfles for example)
Use case
I want to add several lags of a feature. I don't know which one brought more value to my final prediction
sm-ak-r33
commented
4 days ago
Description
As every time we use Exogenous features we need something to explain a posteriori why a feature brings more value than an other one to the final prediction. Feature importance or SHAP are some tools that we have in MLForecasts I would love to have somehing similar in Statsforecast, specially for models that accepts exogenous features (like mfles for example)
Use case
I want to add several lags of a feature. I don't know which one brought more value to my final prediction