Create Decision Tree online

With numiqo you can create a CART Decision Tree from one dependent variable and two or more independent variables. If the dependent variable is categorical, the calculator creates a classification tree. If the dependent variable is metric, it creates a regression tree.

CART Decision Tree Calculator

CART stands for Classification and Regression Trees. The algorithm recursively splits the data into two branches. Metric predictors are split by thresholds, while categorical predictors are split by category groups. For classification trees, numiqo uses Gini impurity. For regression trees, it uses squared error and variance reduction.

The output includes a method table, performance table, classification confusion matrix or regression metrics, relative variable importance, and a compact table of the fitted tree rules.

CART or CHAID

CART is the default Decision Tree method and supports categorical and metric dependent variables. If the dependent variable is categorical, CHAID is also available as a separate method. The CHAID decision tree uses chi-square tests and can create multiple branches, so it is often useful for segmentation of categorical data.

How to interpret a decision tree

The root node contains all rows used in the analysis. Each following row describes a split rule, the number of rows in that node, and the predicted class or predicted metric value. Important variables are those that reduce impurity or error most strongly across the fitted tree.