Tutorials

Six worked analyses, each from raw data to a result you can plot. They follow the examples of Ramsay, Hooker and Graves, Functional Data Analysis with R and MATLAB (2009), so you can compare with the book and with R's fda.

Read the Quickstart first if you have never used fdatools. Every code block runs as written, top to bottom, and ends with an R equivalent section for readers who know R's fda.

Tutorial	You will learn	Data
Smoothing	smooth(): automatic basis, λ by GCV, df=, the harmonic accelerator, monotone curves, SmoothResult	growth, Canadian weather
Functional PCA and CCA	FPCA: harmonics, scores, variance shares, varimax rotation; FCCA	Canadian weather
Registration	landmark and continuous register(), warping functions, amplitude and phase with .decompose()	growth
Functional regression	fregress() with scalar and functional responses, formulas, predict / stderr / cv, the permutation F test	Canadian weather
Dynamics	principal differential analysis with PDA, solve(), the stability diagram, phase-plane plots	lip movement, growth
Machine learning	scikit-learn Pipeline and GridSearchCV; PyTorch BasisLayer, SmoothingLayer, FDataDataset	growth

Data

The growth, gait and pinch datasets ship inside fdatools. The others (Canadian weather, lip, and the rest of the book's datasets) are downloaded once on first use, checked against a SHA-256 checksum, and cached in ~/.cache/fdatools. Set the environment variable FDATOOLS_DATA_DIR to use another folder, for example one you copied to a machine without internet.

Plots

The tutorials draw with matplotlib (the plot extra). The machine learning tutorial also needs PyTorch (the torch extra): uv add "fdatools[plot,torch]".