Skip to content

Gesture ML notebooks: prose/code mismatches (SelectKBest k=4 vs k=6; hard-coded 65.5% with unseeded split) #23

Description

@jonfroehlich

Found in the #16 correctness/prose/comments audit. Small prose/code mismatches in the gesture ML notebooks — text that disagrees with what the code does.

  1. Feature Selection and Hyperparameter Tuning.ipynb — SelectKBest k. The markdown says "Let's try it with k=4." but the code runs SelectKBest(k=6). Make them agree.

  2. GestureRecognizer-FeatureBased.ipynb — hard-coded accuracy. The narrative states a specific number ("Not bad… 65.5% accuracy with just a few simple features"), but the relevant StratifiedKFold(n_splits=5, shuffle=True) is created without a random_state, so the split — and that number — change every run, and the prose will frequently contradict the output. Either seed the split or soften the wording (e.g. "~65%").

Metadata

Metadata

Assignees

No one assigned

    Labels

    documentationImprovements or additions to documentation

    Type

    No type

    Fields

    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions