Methods catalogue

Every method in the library — the PLS / selection algorithms from benchmarks.parity_timing.registry.METHODS and the C++ preprocessing / augmentation / filter / splitter operators from the n4m binding — documented with parameters, bibliographic source, mathematical principle, binding signatures, and benchmark rows.

Total catalogued native methods: 208. Additional Python reference surfaces are documented where relevant.

Core PLS

Method	Description	Tolerance	Refs
cppls	CPPLS (Canonical Powered PLS, Indahl Liland & Næs 2009)	0.1	R
opls	Orthogonal PLS (Trygg & Wold 2002)	0.001	R
pcr	Principal Components Regression	1e-06	Py, R
pls	SIMPLS PLS regression baseline	0.1	Py, R, ikpls, mixOmics
recursive_pls	Recursive (moving-window) PLS	0.1	Py, R

Sparse

Method	Description	Tolerance	Refs
fused_sparse_pls	Fused sparse PLS (§7)	0.05	—
group_sparse_pls	Group sparse PLS (§7)	0.05	Py, R
sparse_pls_da	Sparse PLS-DA (§7)	2.0	Py, R
sparse_simpls	Sparse SIMPLS with soft-threshold lambda	1.0	Py, R

Ensemble

Method	Description	Tolerance	Refs
bagging_pls	Bagging PLS (§20)	1e-06	Py
boosting_pls	Boosting PLS (§20)	1e-06	R
moment_stack	Train-only OOF stack over native moment heads	n/a	n4m
random_subspace_pls	Random-subspace PLS (§20)	1e-06	Py

Robust / weighted

Method	Description	Tolerance	Refs
robust_pls	Robust PLS (Partial Robust M-regression, Serneels 2005)	0.1	R
weighted_pls	Sample-weighted PLS (sqrt(w)-prescaled NIPALS)	0.1	Py

Nonlinear / local

Method	Description	Tolerance	Refs
continuum_regression	Continuum regression (interpolates PLS / OLS)	1e-06	Py, R
gpr_pls	GPR-on-PLS — RBF Gaussian Process on PLS scores (§47)	1e-08	Py
kernel_pls_rbf	Non-linear kernel PLS (RBF kernel)	2.0	R
lw_pls	LW-PLS — Locally-weighted PLS (§17 Phase 4)	5.0	Py

Multi-block / cross-modal

Method	Description	Tolerance	Refs
mb_pls	MB-PLS — Multi-block PLS (§17 Phase 4)	2.0	Py
mir_pls	MIR-PLS — Inverse-regression PLS (§13)	0.05	—
n_pls	N-PLS — 3-way tensor PLS (PARAFAC + OLS by default; Bro 1996 opt-in)	1e-06	Py
o2pls	O2PLS — bi-directional OPLS (Trygg & Wold 2003)	1e-10	R
on_pls	OnPLS — Orthogonal multi-block decomposition (§18)	1e-06	Py
rosa	ROSA — Response-Oriented Sequential Alternation (§19)	1e-06	Py, R
so_pls	SO-PLS — Sequential & Orthogonalized multi-block PLS (§17)	1e-06	R

Calibration transfer

Method	Description	Tolerance	Refs
di_pls	Domain-invariant PLS	1e-06	Py
ds	DS — Direct Standardization (§13)	0.5	R
ecr	Elastic Component Regression (Phase 50)	0.001	Py
pds	PDS — Piecewise Direct Standardization (§13)	0.5	R

Classification & GLM

Method	Description	Tolerance	Refs
pls_cox	PLS-Cox (§5) — Cox PH on PLS scores	1e-06	Py
pls_glm	PLS-GLM (§5) — softmax/Poisson IRLS on PLS scores	1e-06	R
pls_lda	PLS-LDA — LDA on PLS scores (§17 Phase 4)	5.0	Py
pls_logistic	PLS-Logistic — Logistic regression on PLS scores	5.0	Py
pls_qda	PLS-QDA (§5) — quadratic discriminant on PLS scores	1e-06	Py

Missing data

Method	Description	Tolerance	Refs
missing_aware_nipals	Missing-aware NIPALS PLS (§13)	10.0	R

Regularised

Method	Description	Tolerance	Refs
ridge	Direct closed-form Ridge regression	1e-10	Py
ridge_pls	Ridge-augmented PLS	0.1	Py

Diagnostic

Method	Description	Tolerance	Refs
aom_chain_fixed_fit	Final-only reusable fit for one selected strict-linear AOM chain/head/parameter	1e-09	n4m
aom_chain_ridge_pls	Strict-linear AOM single-chain Ridge-PLS selector	1e-08	n4m
aom_chain_screen_refit	Two-pass strict-linear AOM preprocessing screen with exact-CV refit and reusable final model	1e-09	n4m
aom_chain_sweep_run	User-defined native AOM strict-linear chain sweep	1e-09	n4m
aom_pls	Native AOM global strict-linear operator selector for PLS	1e-09	n4m
aom_pls_superblock	Strict-linear AOM-PLS operator superblock	1e-09	n4m
aom_operator_pls_stack	Native AOM strict-operator PLS score stack with Ridge head	1e-09	n4m
aom_staged_chain_campaign	Staged strict-chain AOM/moment preprocessing screen/refit campaign	n/a	n4m
aom_ridge_blender	Native fold-safe AOM-Ridge OOF simplex blender	1e-09	n4m
aom_ridge_active_superblock	Strict-linear AOM Ridge active-superblock selector	1e-09	n4m
aom_ridge_global	Strict-linear AOM Ridge global operator selector	1e-09	n4m
aom_ridge_mkl_superblock	Strict-linear AOM Ridge MKL-light weighted superblock	1e-09	n4m
aom_ridge_pls_superblock	Strict-linear AOM Ridge-PLS operator superblock	1e-09	n4m
aom_ridge_superblock	Strict-linear AOM Ridge superblock over native Ridge fits	1e-09	n4m
aom_robust_hpo	Native strict-linear AOM robust-HPO preprocessing screen	1e-10	n4m
aom_sweep_run	Configurable native AOM strict-linear preprocessing sweep	1e-09	n4m
aom_preprocess	Native AOM operator-bank preprocessing primitive	5.0	n4m
approximate_press	Approximate-PRESS component selection (§29)	1e-10	R
one_se_rule	One-SE component selection rule (§10)	1e-06	R
pls_diagnostic_dmodx	PLS Distance-to-Model X (§9)	5.0	R
pls_diagnostic_q	PLS Q residuals / SPE (§9)	5.0	R
pls_diagnostic_t2	PLS Hotelling T² (§9)	10.0	R
pls_monitoring	PLS process monitoring (T²/Q thresholds + alarms) (§19)	10.0	R
pop_pls	Native POP-PLS per-component strict-linear operator selector	1e-09	n4m

Variable selector

Method	Description	Tolerance	Refs
bipls_select	biPLS backward interval elimination (§18 Phase 5p)	0.7	R
bve_select	Backward Variable Elimination (§18 Phase 5k)	1e-06	R
cars_select	CARS competitive adaptive reweighted sampling	0.0	R
emcuve_select	EMCUVE ensemble MC-UVE (§18 Phase 5n)	1e-06	R
ga_select	GA-PLS genetic algorithm selection	1e-06	R
interval_select	Interval/iPLS forward selection (§18 Phase 5b)	1e-06	R
ipw_select	IPW-PLS iterative predictor weighting (§18 Phase 5t)	1e-06	R
irf_select	Interval Random Frog (Phase 52)	1e-06	Py
iriv_select	Iteratively Retains Informative Variables (Phase 51)	1e-06	Py
pso_select	PSO-PLS — Binary Particle Swarm variable selection (§48)	1e-06	Py
random_frog_select	Random Frog selection (§18 Phase 5g)	1e-06	Py
randomization_select	Randomization-test selector (§18 Phase 5o)	1e-06	R
rep_select	REP-PLS repeated VIP selection (§18 Phase 5s)	1e-06	R
scars_select	SCARS stability + CARS (§18 Phase 5h)	1e-06	Py
shaving_select	Shaving iterative variable trimming	1e-06	R
sipls_select	siPLS synergistic interval selection (§18 Phase 5q)	0.7	—
spa_select	SPA Successive Projections (§18 Phase 5e)	1e-06	R
st_select	ST-PLS soft-thresholded sparse PLS (§18 Phase 5u)	1e-06	R
stability_select	Stability/MCUVE selection (§18 Phase 5c)	1e-06	R
t2_select	T²-PLS loading-weight selection (§18 Phase 5l)	1.2	R
uve_select	UVE noise-thresholded selection (§18 Phase 5d)	1e-06	R
variable_select_coef	|Coef| top-k selection (§18 Phase 5a, method=1)	1.1	R
variable_select_sr	Selectivity-Ratio top-k (§18 Phase 5a, method=2)	1e-06	R
variable_select_vip	VIP top-k variable selection (§18 Phase 5a, method=0)	1e-06	R
vip_spa_select	VIP_SPA — VIP-mask then SPA greedy (Phase 53)	1e-06	Py
vissa_select	VISSA-PLS — Variable Iterative Space Shrinkage (§49)	1e-06	Py
wvc_select	WVC weighted-variable-component top-k	0.7	R
wvc_threshold_select	WVC threshold-based selection (§18 Phase 5r)	1e-06	R

Preprocessing

Operator	Binding	Description
pp_area	n4m.sklearn.AreaNormalization	Per-row area normalisation.
pp_baseline	n4m.sklearn.BaselineCenter	Column-mean baseline centering.
pp_crop	n4m.sklearn.CropTransformer	Slice wavelength columns in the half-open interval [start, end).
pp_derivate	n4m.sklearn.Derivate	Finite-difference derivative along the wavelength axis.
pp_emsc	n4m.sklearn.EMSC	Extended Multiplicative Scatter Correction (polynomial).
pp_first_derivative	n4m.sklearn.FirstDerivative	np.gradient(X, delta, axis=1, edge_order=...) (shape-preserving).
pp_frac_to_pct	n4m.sklearn.FractionToPercent	Convert fractional reflectance/transmittance to percent.
pp_from_absorbance	n4m.sklearn.FromAbsorbance	R = 10**(-A), optionally returned as percent.
pp_gaussian	n4m.sklearn.Gaussian	SciPy-compatible 1-D Gaussian filter along the wavelength axis.
pp_kbins_disc	n4m.sklearn.IntegerKBinsDiscretizer	Per-column integer binning using uniform or quantile edges.
pp_kubelka_munk	n4m.sklearn.KubelkaMunk	KM = (1 - R)^2 / (2 R), with R guarded by epsilon.
pp_log	n4m.sklearn.LogTransform	Element-wise logarithm with optional fit-time auto-offset.
pp_lsnv	n4m.sklearn.LSNV	Sliding-window (local) SNV.
pp_msc	n4m.sklearn.MSC	Multiplicative Scatter Correction.
pp_normalize	n4m.sklearn.Normalize	Column-wise normalisation.
pp_norris_williams	n4m.sklearn.NorrisWilliams	Segment smoothing followed by gap finite differences.
pp_pct_to_frac	n4m.sklearn.PercentToFraction	Convert percent reflectance/transmittance to fraction.
pp_range_disc	n4m.sklearn.RangeDiscretizer	Integer binning against monotonic numeric edges.
pp_resample	n4m.sklearn.ResampleTransformer	Resize spectra to a fixed column count.
pp_resampler	n4m.sklearn.Resampler	Interpolate spectra from a fitted source wavelength grid to a target grid.
pp_rnv	n4m.sklearn.RNV	Robust SNV using median + k * MAD.
pp_savgol	n4m.sklearn.SavitzkyGolay	scipy.signal.savgol_filter parity.
pp_second_derivative	n4m.sklearn.SecondDerivative	Two passes of np.gradient (shape-preserving).
pp_simple_scale	n4m.sklearn.SimpleScale	Column-wise min-max scaling to [0, 1].
pp_snv	n4m.sklearn.SNV	Standard Normal Variate normalisation.
pp_to_absorbance	n4m.sklearn.ToAbsorbance	A = -log10(max(R, epsilon)). Optional %-scaling.

Baseline correction

Operator	Binding	Description
pp_airpls	n4m.sklearn.AirPLS	Adaptive iteratively reweighted PLS (Zhang 2010).
pp_arpls	n4m.sklearn.ArPLS	Asymmetrically reweighted penalized least squares.
pp_asls	n4m.sklearn.AsLS	Asymmetric Least Squares (Eilers & Boelens 2005).
pp_beads	n4m.sklearn.BEADS	Baseline estimation and denoising with sparsity.
pp_detrend	n4m.sklearn.Detrend	Polynomial baseline subtraction.
pp_iasls	n4m.sklearn.IAsLS	Improved asymmetric least-squares baseline correction.
pp_imodpoly	n4m.sklearn.IModPoly	Improved modified polynomial baseline correction.
pp_modpoly	n4m.sklearn.ModPoly	Modified polynomial baseline correction.
pp_rolling_ball	n4m.sklearn.RollingBall	Rolling-ball morphological baseline correction.
pp_snip	n4m.sklearn.SNIP	Statistics-sensitive nonlinear iterative peak-clipping baseline.

Signal transforms

Operator	Binding	Description
filter_correlation	n4m.sklearn.CorrelationFilter	Model-agnostic feature filter by absolute correlation to y.
filter_variance	n4m.sklearn.VarianceFilter	Model-agnostic feature filter by variance.
interval_generator	n4m.sklearn.IntervalGenerator	Generate fixed or overlapping wavelength intervals.
pp_cow_align	n4m.sklearn.CorrelationOptimizedWarping	Segment-wise correlation optimized warping approximation.
pp_direct_standardization	n4m.sklearn.DirectStandardization	Direct standardization transfer map between paired instruments.
pp_dtw_align	n4m.sklearn.DynamicTimeWarpingAlignment	Dynamic-time-warping alignment to a fixed-length reference.
pp_icoshift_align	n4m.sklearn.IcoshiftAlignment	Interval correlation shifting with fixed-size intervals.
pp_local_centering	n4m.sklearn.LocalCentering	Transfer by subtracting source mean and adding target mean.
pp_localized_msc	n4m.sklearn.LocalizedMSC	Feature-wise MSC using a moving local wavelength window.
pp_piecewise_direct_standardization	n4m.sklearn.PiecewiseDirectStandardization	PDS: local regressions mapping source windows to target wavelengths.
pp_piecewise_msc	n4m.sklearn.PiecewiseMSC	Apply MSC independently inside fixed wavelength intervals.
pp_piecewise_snv	n4m.sklearn.PiecewiseSNV	Apply SNV independently inside fixed wavelength intervals.
pp_robust_direct_standardization	n4m.sklearn.RobustDirectStandardization	Direct standardization with iterative residual trimming.
pp_saps	n4m.sklearn.ScoreAugmentedProjectionStandardization	Score-augmented projection standardization inspired by SA-PBS.
pp_slope_bias	n4m.sklearn.SlopeBiasCorrection	Linear slope/bias correction for transferred predictions.
pp_vsn	n4m.sklearn.VariableSortingNormalization	VSN-style data-derived weighted SNV.
pp_weighted_snv	n4m.sklearn.WeightedSNV	Weighted standard normal variate normalization.
pp_xcorr_align	n4m.sklearn.CrossCorrelationAlignment	Whole-spectrum integer shift chosen by maximum correlation.

Feature extraction

Operator	Binding	Description
pp_epo	n4m.sklearn.EPO	External Parameter Orthogonalisation.
pp_fck_static	n4m.sklearn.FCKStaticTransformer	Static fractional convolutional kernel bank transformer.
pp_flex_pca	n4m.sklearn.FlexiblePCA	PCA with integer or explained-variance component selection.
pp_flex_svd	n4m.sklearn.FlexibleSVD	SVD with integer component selection.
pp_osc	n4m.sklearn.OSC	Orthogonal Signal Correction.

Augmentation

Operator	Binding	Description
aug_band_mask	n4m.sklearn.BandMasking	Mask random spectral bands with zero-fill or interpolation.
aug_band_perturb	n4m.sklearn.BandPerturbationAugmenter	Random band-local gain and offset perturbations.
aug_batch_effect	n4m.sklearn.BatchEffectAugmenter	Random offset, slope and gain batch effects.
aug_channel_dropout	n4m.sklearn.ChannelDropout	Randomly drop individual wavelength channels.
aug_dead_band	n4m.sklearn.DeadBandAugmenter	Simulate dead spectral detector bands.
aug_detector_rolloff	n4m.sklearn.DetectorRollOffAugmenter	Detector edge roll-off artifact.
aug_edge_artifacts	n4m.sklearn.EdgeArtifactsAugmenter	Combined edge artifact augmenter.
aug_edge_curve	n4m.sklearn.EdgeCurvatureAugmenter	Curved edge response artifact.
aug_emsc_distort	n4m.sklearn.EMSCDistortionAugmenter	Random EMSC-like multiplicative, additive and polynomial distortion.
aug_gauss_jitter	n4m.sklearn.GaussianJitter	Gaussian smoothing jitter.
aug_gaussian_noise	n4m.sklearn.GaussianAdditiveNoise	Add IID Gaussian noise to each element of X.
aug_hetero_noise	n4m.sklearn.HeteroscedasticNoiseAugmenter	Noise whose standard deviation depends on signal magnitude.
aug_instrument_broaden	n4m.sklearn.InstrumentalBroadeningAugmenter	Instrumental spectral broadening via Gaussian convolution.
aug_linear_drift	n4m.sklearn.LinearBaselineDrift	Add random offset and linear slope drift.
aug_local_clip	n4m.sklearn.LocalClip	Clip random local spectral regions.
aug_local_mixup	n4m.sklearn.LocalMixupAugmenter	Neighbor-constrained mixup augmentation.
aug_local_warp	n4m.sklearn.LocalWarpAugmenter	Random local wavelength warping.
aug_magnitude_warp	n4m.sklearn.MagnitudeWarp	Smooth multiplicative magnitude warp.
aug_mixup	n4m.sklearn.MixupAugmenter	Batch-wise mixup augmentation.
aug_moisture	n4m.sklearn.MoistureAugmenter	Water activity and moisture-content spectral perturbation.
aug_multiplicative_noise	n4m.sklearn.MultiplicativeNoise	Apply per-element multiplicative Gaussian noise.
aug_particle_size	n4m.sklearn.ParticleSizeAugmenter	Particle-size and path-length scattering simulation.
aug_path_length	n4m.sklearn.PathLengthAugmenter	Simulate multiplicative path-length variation.
aug_poly_drift	n4m.sklearn.PolynomialBaselineDrift	Add random polynomial baseline drift.
aug_random_x_op	n4m.sklearn.RandomXOperation	Random element-wise multiply/add/subtract operation.
aug_rotate_translate	n4m.sklearn.RotateTranslateAugmenter	Random rotate/translate spectral augmentation.
aug_scatter_sim	n4m.sklearn.ScatterSimulationMSC	MSC-style multiplicative/additive scatter simulation.
aug_spike_noise	n4m.sklearn.SpikeNoise	Inject random spike artifacts into spectra.
aug_spline_curve_simplify	n4m.sklearn.SplineCurveSimplificationAugmenter	Cubic B-spline refit through a random control subset along the curve.
aug_spline_smooth	n4m.sklearn.SplineSmoothingAugmenter	Spline smoothing augmenter.
aug_spline_x_perturb	n4m.sklearn.SplineXPerturbationAugmenter	Spline x-axis perturbation augmenter.
aug_spline_x_simplify	n4m.sklearn.SplineXSimplificationAugmenter	Cubic B-spline refit through a random control subset on the x-axis.
aug_spline_y_perturb	n4m.sklearn.SplineYPerturbationAugmenter	Spline y-axis perturbation augmenter.
aug_stray_light	n4m.sklearn.StrayLightAugmenter	Stray-light edge artifact.
aug_temperature	n4m.sklearn.TemperatureAugmenter	Temperature-induced shift, intensity and broadening perturbations.
aug_truncated_peak	n4m.sklearn.TruncatedPeakAugmenter	Truncated peaks near spectral edges.
aug_unsharp_mask	n4m.sklearn.UnsharpMask	Random unsharp spectral mask.
aug_wavelength_shift	n4m.sklearn.WavelengthShift	Random spectral shift with linear interpolation.
aug_wavelength_stretch	n4m.sklearn.WavelengthStretch	Random wavelength-axis stretching.
pp_haar	n4m.sklearn.Haar	Single-level Haar DWT coefficient transform.
pp_wavelet	n4m.sklearn.Wavelet	Single-level DWT coefficient transform.
pp_wavelet_denoise	n4m.sklearn.WaveletDenoise	Multi-level DWT VisuShrink denoising.
pp_wavelet_features	n4m.sklearn.WaveletFeatures	Multi-level DWT summary features.
pp_wavelet_pca	n4m.sklearn.WaveletPCA	DWT coefficient projection through PCA scores.
pp_wavelet_svd	n4m.sklearn.WaveletSVD	DWT coefficient projection through SVD scores.

Sample / feature filters

Operator	Binding	Description
filter_composite	n4m.sklearn.CompositeFilter	Boolean composition of leverage and quality filters.
filter_leverage	n4m.sklearn.HighLeverageFilter	Hat-matrix or PCA score-space leverage filter.
filter_quality	n4m.sklearn.SpectralQualityFilter	Stateless row-level spectrum quality filter.
filter_x_outlier	n4m.sklearn.XOutlierFilter	Multivariate outlier filter on the design matrix X.
filter_y_outlier	n4m.sklearn.YOutlierFilter	Univariate outlier filter on the target vector y.

Splitters

Operator	Binding	Description
split_binned_strat_group_kfold	n4m.sklearn.BinnedStratifiedGroupKFoldSplitter	Stratified group k-fold splitter after binning continuous y.
split_kbins_stratified	n4m.sklearn.KBinsStratifiedSplitter	Stratified split using K equal-width or quantile bins of y.
split_kennard_stone	n4m.sklearn.KennardStoneSplitter	Kennard-Stone train/test split.
split_kmeans	n4m.sklearn.KMeansSplitter	K-means++ diversity splitter.
split_split_splitter	n4m.sklearn.SPlitSplitter	SPlit data-twinning splitter.
split_spxy	n4m.sklearn.SPXYSplitter	SPXY (Sample set Partitioning based on X and Y) train/test split.
split_spxy_fold	n4m.sklearn.SPXYFoldSplitter	SPXY k-fold splitter over paired X and y matrices.
split_spxy_g_fold	n4m.sklearn.SPXYGroupFoldSplitter	Group-aware SPXY k-fold splitter.
split_systematic_circular	n4m.sklearn.SystematicCircularSplitter	Systematic circular split over sorted or ordered targets.

Utilities

Method	Binding	Description
moments	n4m.moments	Raw and centered row-additive moment sufficient statistics for exact linear screens.
sweep_run	n4m.sweep_run	Native moment Ridge-CV sweep over candidate lambdas.

---

See the benchmark overview for how parity and timing are measured, and the GitHub Pages dashboard for an interactive cross-method comparison.