3 ms·
It would be helpful if you defined "distribution of minimal depth", described how to interpret those statistics, and why we should care about them. I tried Goog
by psychometry 9y ago
It would be helpful if you defined "distribution of minimal depth", described how to interpret those statistics, and why we should care about them. I tried Googling the term to find out more, but to no avail. As a result, I couldn't follow the rest of the vignette.
- yorwba 9y agoI don't know about Random Forests specifically, but in trees (or forests), "depth" refers to the number of steps you need to take from the root to a specific node. So the "minimal depth" of a variable would be the, well, minimal "depth" of any node in the tree that splits on that variable. As I understand the article, this can be used to compare the importance of different variables, since a more important variable would allow better splits, and appear at a lower depth in the trees.