|
ppforest2 v0.1.0
Projection Pursuit Decision Trees and Random Forests
|
Per-node data collected by routing observations through the tree. More...
#include <Visualization.hpp>
Public Attributes | |
| int | depth |
| std::vector< types::Response > | groups |
| bool | is_leaf |
| std::vector< types::Feature > | projected_values |
| types::FeatureVector | projector |
| types::Feature | threshold |
| types::Response | value |
Per-node data collected by routing observations through the tree.
For internal (condition) nodes:
projected_values: dot product of each reaching observation with the node's projector vector.groups: group label (1-indexed response) of each reaching observation, parallel to projected_values.projector, threshold: the split parameters (copied for convenience so the R side can render histograms without re-accessing the tree).For leaf (response) nodes:
projected_values is empty.groups contains labels of observations reaching the leaf.value is the predicted group (0-indexed response). | int ppforest2::viz::NodeData::depth |
| std::vector<types::Response> ppforest2::viz::NodeData::groups |
| bool ppforest2::viz::NodeData::is_leaf |
| std::vector<types::Feature> ppforest2::viz::NodeData::projected_values |
| types::FeatureVector ppforest2::viz::NodeData::projector |
| types::Feature ppforest2::viz::NodeData::threshold |
| types::Response ppforest2::viz::NodeData::value |