numpypolyfit

Numpypolyfit is a Python library that provides polynomial fitting utilities built on NumPy. It offers a NumPy-friendly interface for fitting polynomial models to one-dimensional data, extending the capabilities of numpy.polyfit with features for uncertainty estimation and diagnostic information.

At its core, the library solves linear least-squares problems to determine polynomial coefficients, with options for weighted fitting and robust numerical methods such as QR decomposition or SVD. It can return a covariance matrix for the coefficients, provides residual diagnostics, and supports basic metrics for assessing goodness of fit. Some implementations also offer conveniences for selecting the degree via information criteria or simple cross-validation.

API and usage: The primary function, polyfit(x, y, degree, w=None, cov=False), returns coefficients and, when cov is True, a covariance matrix. A companion function can evaluate the fitted model on new data. Example usage: import numpypolyfit as npf, numpy as np; x = np.linspace(0, 10, 50); y = 1 + 2*x - 0.5*x**2 + np.random.normal(scale=0.5, size=x.size); coeffs, cov = npf.polyfit(x, y, degree=2, cov=True); model = np.poly1d(coeffs); yfit = model(x).

Numpypolyfit is designed as a lightweight tool for straightforward polynomial regression on univariate data. It is separate from NumPy proper and complements, rather than replaces, more feature-rich libraries such as scikit-learn when broader modeling capabilities are needed.

See also: numpy.polyfit, numpy.poly1d, scikit-learn.