Nixtla's Open Source Time Series Ecosystem.

Nixtla Community

We rely on the models implementations. For example for LightGBM just having the categorical data type in some columns will automatically treat them as categorical. For xgboost you have to <https://xgboost.readthedocs.io/en/stable/tutorials/categorical.html#training-with-scikit-learn-interface|set some arguments>

In <https://www.kaggle.com/code/lemuz90/m5-mlforecast-eval/notebook|this notebook> all the static features are categorical

Interesting, perhaps it is lightgbm giving error. I always thought that lightgbm solved the issue of categorical variables and now should treat them natively. Thanks <@U03UX9KR75W>