Do we need to scale data for LightGBM?

Do we need to scale data for LightGBM?

Generally, in tree-based models the scale of the features does not matter. This is because at each tree level, the score of a possible split will be equal whether the respective feature has been scaled or not.

How can I improve my LightGBM model?

1. Add More Computational Resources.
2. Use a GPU-enabled version of LightGBM.
3. Grow Shallower Trees. Decrease max_depth. Decrease num_leaves.
4. Grow Less Trees. Decrease num_iterations.
5. Consider Fewer Splits. Enable Feature Pre-Filtering When Creating Dataset.
6. Use Less Data. Use Bagging.
7. Save Constructed Datasets with save_binary.

How does LightGBM algorithm work?

While other algorithms trees grow horizontally, LightGBM algorithm grows vertically meaning it grows leaf-wise and other algorithms grow level-wise. LightGBM chooses the leaf with large loss to grow. It can lower down more loss than a level wise algorithm when growing the same leaf.

How does LightGBM work on variables with different scale?

This is documented at http://lightgbm.readthedocs.io/en/latest/Parameters.html?highlight=logloss#metric-parameters I would like to understand how LightGBM works on variables with different scale.

How does the feature fraction work in LightGBM?

Feature fraction or sub_feature deals with column sampling, LightGBM will randomly select a subset of features on each iteration (tree). For example, if you set it to 0.6, LightGBM will select 60% of features before training each tree.

When to use LightGBM to speed up training?

LightGBM will randomly select a subset of features on each iteration (tree) if feature_fraction is smaller than 1.0. For example, if you set it to 0.8, LightGBM will select 80% of features before training each tree can be used to speed up training can be used to deal with over-fitting

When does LightGBM randomly select features on a tree?

LightGBM will randomly select a subset of features on each tree node if feature_fraction_bynode is smaller than 1.0. For example, if you set it to 0.8, LightGBM will select 80% of features at each tree node