API

Baseclasses

FeatureDescriptor(key[, resolve_intervals])	The class FeatureDescriptor abstracts the configuration of features the selection methods are to retrieve from the spectral data.
SpectralSelector(model_hyperparams, n_cv_folds)	Top level base class for all Auswahl selectors.
PointSelector([n_features_to_select, ...])	Base class for feature selection methods that select features independently.
IntervalSelector([n_intervals_to_select, ...])	Base class for feature selection methods that select consecutive chunks (intervals) of features.
Convertible()	Selectors subclassing PointSelector, which provide a global score for each feature, can be made eligible for a PointSelector to IntervalSelector conversion facilitated by PseudoIntervalSelector by inheriting from this class.

Wavelength Point Selection

PointSelector([n_features_to_select, ...])	Base class for feature selection methods that select features independently.
CARS([n_features_to_select, n_cars_runs, ...])	Feature selection with Competitive Adaptive Reweighted Sampling (CARS).
MCUVE([n_features_to_select, n_subsets, ...])	Feature selection with Monte Carlo Uninformative Variable Elimination.
RandomFrog([n_features_to_select, ...])	Feature selection with the Random Frog method.
SPA([n_features_to_select, n_cv_folds, pls, ...])	Feature selection with the Successive Projection Algorithm (SPA).
VIP([n_features_to_select, n_cv_folds, pls, ...])	Feature Selection with Variable Importance in Projection.

Wavelength Interval Selection

IntervalSelector([n_intervals_to_select, ...])	Base class for feature selection methods that select consecutive chunks (intervals) of features.
IPLS([n_intervals_to_select, ...])	Interval selection with Interval Partial Least Squares (iPLS).
FiPLS([n_intervals_to_select, ...])	Feature Selection with Forward interval Partial Least Squares (FiPLS).
BiPLS([n_intervals_to_select, ...])	Feature Selection with Backward interval Partial Least Squares (BiPLS).
IntervalRandomFrog([n_intervals_to_select, ...])	Feature selection with the Interval Random Frog (iRF) method.
PseudoIntervalSelector(selector[, ...])	PseudoIntervalSelector transforms a PointSelector subclassing Convertible into an IntervalSelector.

Utilities

optimize_intervals(n_intervals, ...)	The algorithm calculates the optimal non-overlapping placement of n_intervals of width interval_width into the range of features.
util.get_coef_from_pls(pls)	Retrieves the coef attribute from the PLS model in the shape (n_targets, n_features) without triggering a FutureWarning.

Benchmarking API

Benchmarking

benchmarking.benchmark(data, features, methods)

Function performing benchmarking of Interval- and PointSelector feature selectors across different datasets and different parameterizations of the selectors.

Data Handling

benchmarking.load_data_handler(file_path)	Function to load a pickled instance of DataHandler.
benchmarking.DataHandler(datasets, methods, ...)	Data handling class corralling data generated by the benchmarking of different wavelength selection methods.
benchmarking.util.helpers.Selection([...])	Helper class allowing the handling of lists of different lengths in a single pandas DataFrame.

Plotting

benchmarking.plot_score(pod[, dataset, ...])	Plot regression scores of selectors across different number of selected features as box or bar plot.
benchmarking.plot_stability(pod[, dataset, ...])	Plots the stability score of methods for a given metric.
benchmarking.plot_score_vs_stability(pod[, ...])	Plotting a boxplot for the benchmarked methods displaying
benchmarking.plot_selection(pod, n_features)	Plots the selection probability for features of different selectors.
benchmarking.plot_exec_time(pod[, dataset, ...])	Plots execution times of selectors across different number of features to be selected.

Stability Metrics

benchmarking.util.metrics.StabilityScore(...)	Base class for all stability scores useable by the benchmarking system
benchmarking.util.metrics.PairwiseStabilityScore(...)	The class provides the infrastructure for the introduction of new symmetric and pairwise defined stability metrics.
benchmarking.util.metrics.DengScore([...])	Wraps the calculation of the selection stability score for randomized selection methods, according to Deng et al. [R0e48a2638659-1].
benchmarking.util.metrics.ZucknickScore([...])	Wraps the calculation of the stability score according to Zucknick et al. [R00428e8dc090-1].

Misc

benchmarking.util.helpers.load_data_handler(...)

Function to load a pickled instance of DataHandler.