annotate analysis/calcLoss.m @ 33:74d123779d3b
create analysis script to prune and write the relevant features for each level - including an 80% classification threshold on the tree - perhaps this can be addressed
author |
DaveM |
date |
Wed, 15 Mar 2017 16:33:54 +0000 |
parents |
|
children |
|
rev |
line source |
DaveM@33
|
1 function [l,se] = calcLoss(linkList,featureList, row)
|
DaveM@33
|
2
|
DaveM@33
|
3 classList = traceLinkageToBinary(linkList,row);
|
DaveM@33
|
4 X = data(classList>0,featureList{row,1});
|
DaveM@33
|
5 Y = classList(classList>0);
|
DaveM@33
|
6 [l,se] = loss(featureList{row,3},X,Y);
|
DaveM@33
|
7 end
|