Machine Learning

Decision Trees: Build One Yourself

Click on the scatter plot to place split lines. Watch the decision tree diagram grow, regions color in, and accuracy update after every split.

Splits added: 0/3

Datasets tried: 0/2

Dataset

Class 1

Class 0

Add Split

Click on the plot to preview a split

Tree Metrics

Accuracy0.0%

Splits0

Points40

Root Gini—

Feature Space (click to place X-split)

Colored regions = decision boundaries

Decision Tree Structure

Click on the scatter plot to add splits

Gini Impurity

Measures how mixed a node is. Gini = 0 means all one class (pure). The tree minimizes impurity at each split.

Splitting Criteria

At each node, find the feature and value that most reduces impurity. ID3 uses entropy; CART uses Gini.

Overfitting Risk

Deep trees memorize training data. Pruning, max-depth, and min-samples-per-leaf prevent overfitting.