Definition
A decision tree splits data by a sequence of simple rules ("age > 45 and income < 30 thousand") into ever more homogeneous groups until the leaves hold a dominant class or an average value. The C&RT (CART) and CHAID algorithms handle classification and regression, continuous and categorical inputs and missing values. The result reads like a flowchart.
Data mining & machine learning
The biggest strength is interpretability: the tree shows which variables decide and within which bounds, which physicians, risk managers and process engineers appreciate. The weakness is instability — a small change in data can change the structure — and a tendency to overfit, handled by pruning and cross-validation.
Trees are also the building block of more powerful ensemble methods (random forests, boosted trees) that raise accuracy substantially at the cost of interpretability. For a first understanding of the data and discovery of interactions a simple tree is often the fastest route.
In Statistica
The Data Mining menu offers General Classification and Regression Trees (C&RT), CHAID and Interactive Trees where you steer the splits by hand; the tree is drawn as a clear graph with class shares in the nodes, complemented by a confusion matrix, predictor importance and cross-validation. The finished model can be saved for scoring new data.
Related terms
- Random forests and boostingRandom forests and boosting are ensemble methods that combine hundreds of decision trees into one more accurate model.
- Logistic regressionLogistic regression models the probability of a binary outcome — a patient responds to treatment or not, a customer…
- Cross-validationCross-validation estimates how well a model will perform on new data.
- OverfittingOverfitting happens when a model learns not only the real relationships but also the random noise of the training data…
- Discriminant analysisDiscriminant analysis finds combinations of variables that best separate groups known in advance — healthy from ill…
Knowledgebase guides
FAQ
- C&RT or CHAID?
- C&RT makes binary splits and also suits regression; CHAID splits into several groups by a chi-square test and is popular in marketing segmentation with categorical data.
- How do I keep a tree from overfitting?
- Limit the minimum node size and depth, prune by cross-validation and verify accuracy on test data.
Try it on your own data
Statistica free for 30 days
Full version, no credit card. Or get a pricing estimate in a minute.
Data mining & machine learning
Updated: September 2026.