#polynomial #regression #model #numerical-methods #statistics

polyfit-residuals

Efficiently calculate the residual errors (in the least squares sense) of all polynomial models (up to some degree) for a given dataset and compute least-squares polynomial fits

14 releases

0.6.4 Oct 26, 2024
0.6.3 Jun 3, 2024
0.6.1 May 5, 2023
0.5.0 May 2, 2023
0.1.0 Dec 3, 2022

#315 in Math


Used in pcw_regrs

MIT/Apache

57KB
947 lines

Description

This package allows to efficiently compute the residual errors (in the least squares sense) of all the possible polynomial models on all (or all the ones starting at some point) subsegments of some data. So it minimizes ∑ᵢ (P(xᵢ)-yᵢ)² where i ∈ I and I is a subsegment of 1,...,n where the input data is given by x₁,...,xₙ, y₁,...,yₙ; without actually computing the polynomial P. There is also a function to calculate P explicitly for the given data and some particular degree - however this functionality is not considered the primary focus of this package.

Weighted least squares is also supported. Related methods may be found in the weighted module.

Algorithm

The algorithm used is essentially a Givens rotation based QR decomposition. For numerical reasons it uses a Newton basis throughout which should yield better results than some other implementations based on the monomial basis / Vandermonde matrix due to being better conditioned. This might incur some additional runtime costs in some places and save some in others. The algorithm for all segments starting at 0 should be O(nd) and the one for all segments should be O(n²d²) where n is the size of the input and d the maximal polynomial degree.

Performance

Criterion benchmarks run on the author's machine (AMD Ryzen 9 5900X @ ~4.6GHz) may be found here.

Dependencies

~2.5MB
~43K SLoC