During model interpretation, if you see an interesting leaf/node, a very useful functionality would be to see the raw samples from that particular node. Having this samples in a dataframe, you can simple browse through them to check theirs similarities/characteristics or apply any kind of analysis that you need.
parrt
commented
1 year ago
During model interpretation, if you see an interesting leaf/node, a very useful functionality would be to see the raw samples from that particular node. Having this samples in a dataframe, you can simple browse through them to check theirs similarities/characteristics or apply any kind of analysis that you need.