Decision Trees and Ensembles · seed 1 · A4, ink-friendly. The answer key prints on its own page for grown-ups.

One question at a time, then a forest vote

Computing · Machine Learning · ages 20-22
Name ______________________   Date ____________

  1. Which group has higher entropy?

    • Only apples, fully pure
    • Only bananas, fully pure
    • Half apples and half bananas

  2. How does a decision tree sort data?

    • By asking one yes or no question at a time down branches
    • By guessing the group without asking anything
    • By sorting numbers from large to small only

  3. Changing a few training rows can make a single tree grow a different shape with different answers.

    Circle one:   True   False

  4. In a forest, how is each tree fed its training data?

    • Each tree gets an identical copy of all rows
    • Each tree trains on a resample drawn at random with replacement
    • Each tree trains on a single row only

  5. For the first split, the algorithm tries every question. Which one does it pick?

    • The one that lowers entropy the most
    • The one asked least often before
    • The one with the longest wording

  6. How does a random forest turn many tree opinions into one prediction?

    • It always trusts the tallest tree
    • It averages the question wordings
    • Every tree votes and the majority wins

  7. What does a forest gain over one tree, and what does it give up?

    • It gains accuracy and steadiness but gives up readability
    • It gains speed and gives up accuracy
    • It gains readability and gives up voting

  8. Three changed rows reshape your whole tree and flip its answers. What should you conclude?

    • The tree shows healthy stability
    • The tree shows single-tree instability, a forest would be steadier
    • Entropy was computed wrongly
LightMySky · lightmysky.comW1-mt_xo9EEd4R1i-s1

Answer key

For grown-ups. Fold this page away before handing over the rest.

One question at a time, then a forest vote W1-mt_xo9EEd4R1i-s1

  1. Half apples and half bananas · Entropy measures mixed-up-ness, so the half and half group scores highest.
  2. By asking one yes or no question at a time down branches · Each answer picks a branch until the item lands in a final group.
  3. True · Single trees are twitchy: small data shifts can regrow the whole tree.
  4. Each tree trains on a resample drawn at random with replacement · Bootstrapping gives every tree a slightly different dataset to grow from.
  5. The one that lowers entropy the most · The biggest entropy drop makes the purest new groups.
  6. Every tree votes and the majority wins · Majority vote smooths out the odd choices of any single tree.
  7. It gains accuracy and steadiness but gives up readability · Many averaged trees beat one readable tree on accuracy while becoming harder to read.
  8. The tree shows single-tree instability, a forest would be steadier · A full reshape from a tiny data shift is exactly the twitch one tree suffers.
Worksheet · LightMySky