lagodiuk / decision-tree-js Goto Github PK

Not really an issue, but it would be cool if passed an array (say of keywords) we could check against all keywords instead of just pushing the array together and checking for exact matches.

So right now if I pass ['here','are',keywords'] it will just check for 'here,are,keywords'. Which essentially checks for an exact match on the array, but I am looking to predict an attribute, based on how many keywords I match. and the keywords could be variable, so one item may have 1 keyword while another may have 6.

I tried figuring out where I want to modify the js to do this, but I am VERY new to machine learning and so I don't fully understand how everything works together yet.

Please change this line to:
predicate = predicates[tree.predicateName];

This single line change enables using a pre-saved root by setting decisionTree.root = savedRoot

Would be nice to add additional functionality to save the root as well as build a dt object without input data.