Section 0 — Step Narrator (LIVE)
Ready
Click Grow Tree. You will see: leaf selection → candidate scan → best split → branches grow → cards flow.
Tip: Click any card in the arena to highlight its decision path.
Section 1 — What you will learn
✓ By the end you will understand
- Node = a yes/no question the AI asks.
- Split = send items left vs right.
- Gini / Entropy = how mixed a node is.
- Information Gain = how much a question reduced the mess.
- Depth & Min split = tree size limits.
- Pruning = cut weak branches to generalize better.
Section 2 — Controls
Story helps kids remember what the tree is doing.
Higher depth can overfit (memorize noise).
Bigger min split keeps the tree simpler.
Pruning fights overfitting.
Noise = messy real life. Perfect data never exists.
Section 3 — Live stats
Train Accuracy
—
Test Accuracy
—
Nodes
1
Leaves
1
Avg Leaf Purity
—
Generalization Gap
—
Live Chart
Train/Test accuracy trend
Section 4 — Live math (updates during scan)
Press “Grow Tree”. During candidate scan, math + gain will update live.
🧠 Kid Coach
This lab is tree-first. The tree is the product. No dot battlefield.
Try: Grow 3 times → then increase depth → then Break it → then Prune.