Open motiwari opened 2 years ago
Just to be clear, only the boosting part of XGBoost is necessary. We don't need any of the systems-level optimization, just the algorithm
When I implement XGBoost with their package, it runs all the systems-level optimizations by default.
Based on discussions with Martin, this is now low pri in favor of #135 and #134
We should set this up as a baseline.
It may be valuable to have this as a separate class that inherits
Tree
orTreeClassifier
.It should take parameters like the boosting learning rate etc.