API Reference

The RFFLearn library consists of the several sub modules. Click the module name to see the details of each module.

Sub module name	Description
rfflearn.cpu	Regressors and classifiers of random Fourier features on CPU.
rfflearn.gpu	Regressors and classifiers of random Fourier features on GPU.

rfflearn.cpu

This sub-module is a module for machine learning algorithms of random Fourier feature runnable on CPU. This sub-module contains machine learning algorithms (e.g. regressors, classifiers) that are designed to be run on CPU. The interfaces of classes and functions in this module are quite close to the scikit-learn library. Most (but not all) of the classes use the scikit-learn as a backend.

This sub-module contains three types of classes and functions: (1) ML models such as support vector machines with random Fourier features, (2) hyperparameter tuners that are wrapper functions of Optuna, and (3) ML model explainers that are wrapper functions of explainer functions in scikit-learn and SHAP.

ML class	Description
rfflearn.cpu.RFFCCA	Canonical correlation analysis with random Fourier features.
rfflearn.cpu.RFFGPC	Gaussian process classification with random Fourier features.
rfflearn.cpu.RFFGPR	Gaussian process regression with random Fourier features.
rfflearn.cpu.RFFPCA	Principal component analysis with random Fourier features.
rfflearn.cpu.RFFRegression	Regression with random Fourier features.
rfflearn.cpu.RFFSVC	Support vector classification with random Fourier features.
rfflearn.cpu.RFFSVR	Support vector regression with random Fourier features.
rfflearn.cpu.ORF*	ORF (orthogonal random features) version of estimators. For example, `rfflearn.cpu.ORFSVC` is a support vector classifier with ORF. The arguments of constructor and member functions are the same as RFF version, so please see the document of `rfflearn.cpu.RFF*` for the details of the usage of each class.
rfflearn.cpu.QRF*	QRF (quasi-random features) version of estimators. For example, `rfflearn.cpu.QRFSVC` is a support vector classifier with QRF. The arguments of constructor and member functions are the same as RFF version, so please see the document of `rfflearn.cpu.RFF*` for the details of the usage of each class.

Explainer function	Description
rfflearn.cpu.permutation_feature_importance	Calculate permutation importance, and set the feature importance (mean of permutation importance for each trial) as `model.feature_importances_`.
rfflearn.cpu.permutation_plot	Visualize permutation importance as a box diagram.
rfflearn.cpu.shap_feature_importance	Calculate SHAP values using shap library, and set the feature importance (absolute of SHAP values) as `model.feature_importances_`.
rfflearn.cpu.shap_plot	Create a bar plot of SHAP values.

Tuner function	Description
rfflearn.cpu.RFFRegressor_tuner	Hyperparameter tuner for `RFFRegressor` models.
rfflearn.cpu.ORFRegressor_tuner	Hyperparameter tuner for `OFFRegressor` models.
rfflearn.cpu.QRFRegressor_tuner	Hyperparameter tuner for `QFFRegressor` models.
rfflearn.cpu.RFFSVC_tuner	Hyperparameter tuner for `RFFSVC` models.
rfflearn.cpu.ORFSVC_tuner	Hyperparameter tuner for `ORFSVC` models.
rfflearn.cpu.QRFSVC_tuner	Hyperparameter tuner for `QRFSVC` models.
rfflearn.cpu.RFFGPC_tuner	Hyperparameter tuner for `RFFGPC` models.
rfflearn.cpu.ORFGPC_tuner	Hyperparameter tuner for `ORFGPC` models.
rfflearn.cpu.QRFGPC_tuner	Hyperparameter tuner for `QRFGPC` models.
rfflearn.cpu.RFFGPR_tuner	Hyperparameter tuner for `RFFGPR` models.
rfflearn.cpu.ORFGPR_tuner	Hyperparameter tuner for `ORFGPR` models.
rfflearn.cpu.QRFGPR_tuner	Hyperparameter tuner for `QRFGPR` models.

rfflearn.gpu

Sub module for regressors and classifiers of random Fourier feature runnable on GPU. This module is designed to have the same interface as rfflearn.cpu. See the API reference of rfflearn.cpu for the details of this module.