GlueRatGlobal Robotics · machine-learning laboratory

Shape the curve.

A polynomial model is still linear regression: each coefficient has a linear weight, even when the features run from x to x⁵. Turn the six knobs and fit the noisy observations.

← GRG Robotics
Drag ↑↓ · scroll · arrow keys
Your model
Manual fit
ŷ = 0.00 + 0.00x + 0.00x² + 0.00x³ + 0.00x⁴ + 0.00x⁵
Fit quality
Start turning
Six coefficients · fifth-degree polynomial

The y-crossing lifts the whole model; each additional power changes its shape more strongly toward the edges.

Drag ↑↓ · scroll · arrow keys
Try this: hide the observations, then turn one coefficient at a time. β₀ sets the y-crossing; higher powers reshape the ends most strongly.
Observed data vs your curve
data prediction residual
Mean squared error—
R² on these points—
Largest residual · highlighted above—
Polynomial degree: 5 · 36 observations · Gaussian noiseThe model is linear in β, not necessarily straight in x.