Skip to content
Tim Erickson edited this page Sep 28, 2018 · 1 revision

Arbor

This is the "tree tool" designed in collaboration with Joachim Engel and partly inspired by work with Laura Martignon. It lets the user make trees (either classification or regression trees) by hand rather than by using algorithms such as CART or random forest. User make trees by dragging attributes into the trees to make new nodes, and "configuring" the attributes to specify what values of the attributes correspond to which branch at that new node.

The tool reports basic statistics about the tree's performance (e.g., number of false positives) and even emits these (into CODAP) so that users can make their own measures of tree performance, that is, ways to tell whether one tree is better than another.

The tree interacts deeply with CODAP selection.

Clone this wiki locally