Abhimanyu Dayal

Written in context of #1319, DFFML could build a centralised module for implementation of datasets. That is, at present, in order to implement a new dataset, we need to write...

Added gradient boosting binding

5

I noticed that mlpack doesn't currently have a binding for Gradient Boosting. To give a brief description of Gradient Boosting: it's an ensemble technique which uses weak learners such as...

[WIP] XGBoost

3

XGBoost (eXtreme Gradient Boosting) is an optimized implementation of Gradient Boosting. Working forward from PR #3735 where I implement Gradient Boosting.

Started working on new loss functions more specific to XGB. Continuing from PR #3747

Shifted sse loss to decision tree to streamline xgboost functionality and ease regularisation.

5

SSE_loss existed in xgboost directory beforehand. I shifted that to the Decision Trees directory and included that as a gain function to be used. All the xgboost loss functions are...

[WIP] Gradient Boosting binding

2

Continuing from PR #3725 Needed to close that one because of some branch issues.

Cherry-picked from PR #3736 After training, xgboost can calculate feature importance to understand the contribution of each feature in the classification decision. XGBoost provides two main types of feature importance...

Tree Pruning

1

Adding tree pruning method in decision trees as part of the xgboost implementation (PR #3736).