📏 Regression Leverage and Cook's Distance: How One Point Drags the Least-Squares Line - 3D Blender Lab
Hang a bar on one spring per data point and watch it settle on the least-squares line, then move a single suspect point: leverage, standardised residual and Cook's distance show why a far-away point can dominate a regression even when its residual is small.
⚙ Under the hood
Hang a bar on one spring per data point and watch it settle on the least-squares line, then move a single suspect point: leverage, standardised residual and Cook's distance show why a far-away point can dominate a regression even when its residual is small.
regression diagnosticsleverageCook's distanceinfluential pointsleast squareshat matrixresidualsstatistics
3D · Three.js / WebGL renderer · 60 FPS target · runs fully client-side, no install