Statistics Toolbox

treefit

Fit a tree-based model for classification or regression

Syntax

• T = treefit(X,y)
  T = treefit(X,y,'param1',val1,'param2',val2,...)
  

Description

T = treefit(X,y) creates a decision tree T for predicting response y as a function of predictors X. X is an n-by-m matrix of predictor values. y is either a vector of n response values (for regression), or a character array or cell array of strings containing n class names (for classification). Either way, T is a binary tree where each non-terminal node is split based on the values of a column of X.

T = treefit(X,y,'param1',val1,'param2',val2,...) specifies optional parameter name-value pairs. Valid parameters are

For all trees:

'catidx'	Vector of indices of the columns of X. treefit treats these columns as unordered categorical values.
'method'	Either 'classification' (default if y is text) or 'regression' (default if y is numeric)
'splitmin'	A number n such that impure nodes must have n or more observations to be split (default 10)
'prune'	'on' (default) to compute the full tree and a sequence of pruned subtrees, or 'off' for the full tree without pruning

For classification trees only:

'cost'	p-by-p matrix C, where C(i,j) is the cost of classifying a point into class i if its true class is j (default has C(i,j)=1 if i~=j, and C(i,j)=0 if i=j). C can also be a structure S with two fields : S.group containing the group names, and S.cost containing a matrix of cost values.
'splitcriterion'	Criterion for choosing a split: either 'gdi' (default) for Gini's diversity index, 'twoing' for the twoing rule, or 'deviance' for maximum deviance reduction
'priorprob'	Prior probabilities for each class, specified as a vector (one value for each distinct group name) or as a structure S with two fields: S.group containing the group names, and S.prob containing a vector of corresponding probabilities

Examples

Create a classification tree for Fisher's iris data.

•    load fisheriris;
     t = treefit(meas,species);
     treedisp(t,'names',{'SL' 'SW' 'PL' 'PW'});
  
  
  

See Also

treedisp, treetest

References

[1]  Breiman, et al., Classification and Regression Trees, Chapman and Hall, Boca Raton, 1993.

treedisp

treeprune