scikitfda

scikit-fda is an open-source Python library for functional data analysis (FDA). It provides data structures and algorithms for representing and analyzing curves, surfaces, and other functional data, with a design that complements NumPy, SciPy, and scikit-learn.

Two central data representations are FDataGrid, which stores irregular or regular discretizations of curves on a grid, and FDataBasis, in which curves are represented through basis expansions (for example, B-splines via BSplineBasis, Fourier basis, or polynomial bases). The library also includes basis functions, smoothing methods, and derivative/integral operators.

Functional data analysis capabilities include smoothing and denoising, registration (curve alignment), functional principal component analysis (FPCA), and functional regression (scalar-on-function, function-on-scalar, and function-on-function approaches). It also supports functional clustering and classification, as well as functional discriminant analysis.

scikit-fda emphasizes compatibility with scikit-learn, offering estimators and transformers that can be used in pipelines and cross-validation. The project provides extensive documentation, tutorials, and examples, and is actively developed on GitHub with community contributions. It supports Python environments common in data science workflows and integrates with NumPy/SciPy arrays.