Module fit

source code

Least-squares polynomial fitting

Functions

pfit(x, y, order=1)
Fit a polynomial to the scatter plot (x,y), returning yhat and fitted polynomial.

source code

kfit(x, y, kernel, l=1.0, **kargs)
Use the kernel trick to fit a non-parametric estimator.

source code

penfit(x, y, kernel, knots, l=1, **kargs)
Use a kernel and a kernel penalty, but don't use all the x's as knots.

source code

Function Details

Use the kernel trick to fit a non-parametric estimator.

kargs: keyword arguments passed to kernel

Use a kernel and a kernel penalty, but don't use all the x's as knots.

When the kernel corresponds to cubic splines, this technique is sometimes called (penalized) regression splines.