Welcome to GBDTMO’s documentation!

GBDTMO is a gradient boosted decision tree method which supports learning multiple outputs within a single tree. GBDTMO constructs the predicts of all outputs or a subset of automatically selected outputs on a leaf. Compared with GBDT for single output, GBDTMO has better generalization ability and faster training speed. See our paper for technical details.

Contents:

• Python API
  • load_lib
  • create_graph
  • GBDTMulti
  • GBDTSingle
• Parameters
  • Meta
  • Tree
  • Learning
• Examples
  • Plotting tree
  • Using GBDTMO
  • Custom loss

Feel free to contact with us if you have any questions or suggestions.