Statistics Toolbox

treeval

Compute fitted value for decision tree applied to data

Syntax

• YFIT = treeval(T,X)
  YFIT = treeval(T,X,subtrees)
  [YFIT,NODE] = treeval(...)
  [YFIT,NODE,CNAME] = treeval(...)
  

Description

YFIT = treeval(T,X) takes a classification or regression tree T as produced by the treefit function, and a matrix X of predictor values, and produces a vector YFIT of predicted response values. For a regression tree, YFIT(j) is the fitted response value for a point having the predictor values X(j,:). For a classification tree, YFIT(j) is the class number into which the tree would assign the point with data X(j,:). To convert the number into a class name, use the third output argument, cname (below).

YFIT = treeval(T,X,subtrees) takes an additional vector subtrees of pruning levels, with 0 representing the full, unpruned tree. T must include a pruning sequence as created by the treefit or prunetree function. If subtree has k elements and X has n rows, then the output YFIT is an n-by-k matrix, with the jth column containing the fitted values produced by the subtrees(j) subtree. subtrees must be sorted in ascending order.

[YFIT,NODE] = treeval(...) also returns an array NODE of the same size as YFIT containing the node number assigned to each row of X. The treedisp function can display the node numbers for any node you select.

[YFIT,NODE,CNAME] = treeval(...) is valid only for classification trees. It retuns a cell array CNAME containing the predicted class names.

Examples

Find the predicted classifications for Fisher's iris data.

• load fisheriris;
  t = treefit(meas,species);   % Create decision tree
  sfit = treeval(t,meas);      % Find assigned class numbers
  sfit = t.classname(sfit);    % Get class names
  mean(strcmp(sfit,species))   % Compute proportion correctly
                               % classified
  ans =
        0.9800
  

See Also

treefit, treeprune, treetest

treetest

trimmean