How to output decision tree data in sklearn

Question

Is there a way to output decision tree data, eg as a dictionary? I'm aware of sklearn.tree.export_graphviz , which is a great visual, but I was wondering if there was a way to output, say, this:

as this:

{0: 'samples': 100.0%, 'value': [0.53, 0.47],
 1: 'samples': 72.6%, 'value': [0.61, 0.39],
 2: 'samples': 27.4%, 'value': [0.38, 0.62]}

Having the data as a dictionary , DataFrame , array , etc. would make it easier to analyze, rather than just looking at some colored boxes. I've looked through the sklearn documentation but couldn't find anything.

Thanks

Answer 1

It looks like there is an implementation for this here . After playing with it, I think one of the limiting factors is the depth of a dict like this. Remember that a tree like this isn't:

{node1:[data],
 node2:[data],
 node3:[data],
 etc}

It's more like:

{node1:[data],
      sub-node1:{
                 sub-sub-node1:{
                                sub-sub-sub-node1:[data],sub-sub-sub-node2:[data],...}
                 sub-sub-node2:{
                                sub-sub-sub-node1:[data],sub-sub-sub-node2:[data],...}
                 etc}
      sub-node2:{etc}
}

Which can quickly get out of hand if you have a tree with high (or None ) max_depth