In case the feature set if very large, it's hard to monitor the stats generated by the node_stats() method.
One solution would be to have the option to get the stats for the most important 10, 20% of the features.
Another one would be to show only the features belonging to root-leaf path.
This functionality could be useful for any visualisation which handle features, like instance_feature_importance()
In case the feature set if very large, it's hard to monitor the stats generated by the node_stats() method.
One solution would be to have the option to get the stats for the most important 10, 20% of the features. Another one would be to show only the features belonging to root-leaf path.
This functionality could be useful for any visualisation which handle features, like instance_feature_importance()