Teaching a Random Forest to Tell Walking from Running | Medium

Jun 3

Teaching a Random Forest to Tell Walking from Running: A Computer Vision Pipeline with Hand-Built Features Informed by SHAP

How a 56-feature baseline became a 240-feature classifier at 86% accuracy, and what per-class SHAP attribution surfaces about a model's actual confusions.

Watch what happens when you take 250 frames of someone waving their arms and compress them into a single grayscale image where bright means just-now and dim means a moment ago... Continue reading on Medium

Michelle Griffith

Teaching a Random Forest to Tell Walking from Running | Medium

Teaching a Random Forest to Tell Walking from Running: A Computer Vision Pipeline with Hand-Built Features Informed by SHAP

How a 56-feature baseline became a 240-feature classifier at 86% accuracy, and what per-class SHAP attribution surfaces about a model's actual confusions.

Does your step counter undercount women? A Bayesian Audit of a Frequentist Failure to Reject Null Hypothesis

Let the User Edit: How a Static Dashboard Became a Full-Stack Platform (and What Broke Along the Way)