Computing & Software

Decision Tree Split Information Gain Calculator

Evaluate a two-child classification split using parent/child class-count histograms.

How the calculation works

Parent class entropy = shannon_entropy(parent) bits. Weighted child entropy = tree_split_entropy(parent,left,right) bits. Information gain = shannon_entropy(parent)-tree_split_entropy(parent,left,right) bits.

Worked example

Example inputs — Parent class counts: 20, 20; Left-child class counts in the same class order: 20, 0; Right-child class counts in the same class order: 0, 20. Expected results — Parent class entropy: 1 bits; Weighted child entropy: 0 bits; Information gain: 1 bits.

Assumptions and limits

Analytical software/hardware scenario, not a benchmark or a live provider quote. Use the actual architecture, workload and billing definitions; overheads not named in the model are excluded. Every class count must split exactly between the two nonempty children. This scores one candidate split and does not train a tree, penalize overfitting or account for sample weights.

Reference and further reading

Read the underlying subject reference. Calculith is an independent implementation; the reference does not endorse or certify this tool.

Related calculators

Browse all 1000 calculators →

Example values are editable. Results are estimates where assumptions apply. Calculations run in your browser. Updated 2026-11-16.