Tips on LightGBM

Things on this page are fragmentary and immature notes/thoughts of the author. Please read with your own judgement!

It is strongly suggested that you load data into a pandas DataFrame and handle categorical variables by specifying a dtype of "category" for those categorical variables.
```
df.cat_var = df.cat_var.astype("category")
```
This is the easiest way to handle categorical variables in LightGBM. For more details, please refer to Handle Categorical Variables in LightGBM .
The sklearn wrapper of LightGBM lag behind the development of sklearn. Be aware of the latest supported version of sklearn when you use sklearn wrapper of LightGBM. It is suggested that you use the original API of LightGBM to avoid version issues.
It is suggested that you always specify an validation dataset when you train a model using the function train.
LightGBM supports distributed training on multiple machines (without Spark).

https://github.com/microsoft/LightGBM/tree/master/examples/parallel_learning

Hyper Parameter Tuning

Optuna is a good framework for tuning hyper parameters.

https://sites.google.com/view/lauraepp/parameters

https://lightgbm.readthedocs.io/en/latest/Parameters.html

https://lightgbm.readthedocs.io/en/latest/GPU-Tutorial.html

https://lightgbm.readthedocs.io/en/latest/pythonapi/lightgbm.Dataset.html#lightgbm-dataset

https://lightgbm.readthedocs.io/en/latest/_modules/lightgbm/sklearn.html

https://www.kaggle.com/nicapotato/multi-class-lgbm-cv-and-seed-diversification

https://sefiks.com/2018/10/13/a-gentle-introduction-to-lightgbm-for-applied-machine-learning/

https://github.com/microsoft/LightGBM/blob/master/examples/python-guide/sklearn_example.py

https://github.com/microsoft/LightGBM/tree/master/examples/python-guide

https://www.kaggle.com/tapioca/multiclass-lightgbm

https://lightgbm.readthedocs.io/en/latest/Features.html

https://lightgbm.readthedocs.io/en/latest/Parameters-Tuning.html

https://lightgbm.readthedocs.io/en/latest/pythonapi/lightgbm.train.html