mir_pls — MIR-PLS (Mid-InfraRed PLS, regularised)

Group: Multi-block / cross-modal · Registry tolerance: 0.05 · Parity reference: paper-only (Sjöblom, J., Svensson, O., Josefson, M., Kullberg, H. & Wold, S. (1998). An evaluation of orthogonal signal correction applied to calibration transfer of near infrared spectra. Chemom. Intell. Lab. Syst. 44(1-2), 229-244. (Inverse-regression PLS variant; no widely installable port.))

Description

MIR-PLS — Inverse-regression PLS (§13)

From the pls4all.sklearn.MIRPLSRegression docstring:

Multivariate Inverse Regression PLS (Sjöblom 1998).

Parameters

Name	Type	Default	Notes
n_components	int	2	Number of latent components extracted (k).

Explanations

Bibliographic source

Sjöblom, J., Svensson, O., Josefson, M., Kullberg, H. & Wold, S. (1998). An evaluation of orthogonal signal correction applied to calibration transfer of near infrared spectra. Chemometrics and Intelligent Laboratory Systems 44(1–2), 229–244. — adapted for MIR regularisation conventions.

Mathematical principle

MIR-PLS is a PLS variant tuned to mid-infrared spectroscopy conventions: it operates in absorbance space, uses a different ridge constant on the inner regression, and exposes a coefficient-export path that matches the mid-IR community’s expectations (sign and ordering of loadings agree with plsregress rather than pls::plsr).

Algorithmically it is a SIMPLS variant with a fixed ridge constant in the inner regression and an alternate intercept derivation. The differences from plain PLS are numerical, not algorithmic — the resulting predictions are within FP noise of plain PLS for well-conditioned inputs but more stable on the highly-correlated columns typical of MIR FTIR spectra.

Implementation

n4m_mir_pls_fit. No widely installable library reference; treated as paper_only in the registry.

MATLAB header (bindings/matlab/+pls4all/MirRegression.m):

pls4all.MirRegression — Multivariate Inverse Regression PLS.

Usage

Every pls4all binding tab dispatches into the same C kernel; the external libraries listed at the bottom of the page are the parity references registered in benchmarks.parity_timing.registry. Switch tabs to read the same fit in your language. The R package now ships drop-in-compatible facades for the CRAN pls package (plsr, pcr, mvr) and for the mdatools::pls(x, y, ...) matrix idiom — those tabs appear only on the methods that have a meaningful equivalence.

pls4all bindings

C ABI · libn4m

/* C ABI — libn4m */
n4m_context_t* ctx = n4m_context_create();
n4m_config_t*  cfg = n4m_config_create();
n4m_method_result_t* res = NULL;
n4m_mir_pls_fit(ctx, cfg, &x_view, &y_view, /* hyperparams */, &res);
/* … read coefficients / mask / scores via */
/* n4m_method_result_get_double_matrix / vector / scalar … */
n4m_method_result_destroy(res);
n4m_config_destroy(cfg);
n4m_context_destroy(ctx);

Python · pls4all (raw)

import pls4all
from pls4all._methods import mir_pls_fit
with pls4all.Context() as ctx, pls4all.Config() as cfg:
    res = mir_pls_fit(ctx, cfg, X, y, n_components=4)
# then: res.matrix("predictions"), res.matrix("coefficients"),
# res.vector("mask"), res.scalar("intercept"), …

Python · pls4all.sklearn

from pls4all.sklearn import MIRPLSRegression
mdl = MIRPLSRegression(n_components=2)
mdl.fit(X, y)
y_hat = mdl.predict(X_test)

R · pls4all_method()

library(pls4all)
# Unified low-level dispatcher (May 2026 R cleanup):
res <- pls4all_method("mir_pls", X, y,
                      n_components = 4L)
# res is a named list with MethodResult arrays/scalars.
# selected_indices / top_k_intervals are 1-based.

MATLAB · pls4all (MEX)

res = pls4all.mir_pls(X, y, 4);
% see header of bindings/matlab/+pls4all/mir_pls.m for full
% parameter surface:
%   res = mir_pls(X, Y, n_components)
yhat = predict(res, Xtest);

MATLAB · pls4all (classdef)

mdl  = pls4all.fit("mir_pls", X, y, "NumComponents", 4);
yhat = predict(mdl, Xtest);

Registry parity references 📐

• 📜 Paper-only — no executable parity reference; the pls4all implementation is verified by a smoke fit only. Canonical citation: Sjöblom, J., Svensson, O., Josefson, M., Kullberg, H. & Wold, S. (1998). An evaluation of orthogonal signal correction applied to calibration transfer of near infrared spectra. Chemom. Intell. Lab. Syst. 44(1-2), 229-244. (Inverse-regression PLS variant; no widely installable port.)

Benchmarks

Adaptive wall-clock per cell measured against full_matrix.csv. Only backends that implement this method are listed; libraries without the method are omitted.

Verdict  ·  ✓ ref / ≈ ref / ~ shape mark a reference-gate pass at strict / relaxed / qualitative tolerance  ·  ✓ bind = pls4all binding agrees with the C++ baseline  ·  ⇄ cross-check = documented by-design selector/RNG/model, noncanonical API/facade convention, or secondary oracle  ·  ✗ divergent  ·  ⚠ error  ·  — not run. The fastest backend per column is marked 🏆.

1 thread

Backend	Parity	200×30 (ms)
Python · pls4all
pls4all.python	✓ bind	1.27 ms🏆
pls4all.sklearn	✓ 3e-15	1.47 ms
R · pls4all
pls4all.R	✓ bind	3.98 ms
pls4all.R.formula	✓ bind	4.72 ms
pls4all.R.mdatools	✓ bind	5.11 ms
pls4all.R.pls	✓ bind	5.04 ms

3 threads

Backend	Parity	200×30 (ms)
Python · pls4all
pls4all.python	✓ bind	1.29 ms🏆
pls4all.sklearn	✓ 3e-15	1.43 ms
R · pls4all
pls4all.R	✓ bind	4.49 ms
pls4all.R.formula	✓ bind	5.07 ms
pls4all.R.mdatools	✓ bind	5.39 ms
pls4all.R.pls	✓ bind	4.94 ms

10 threads

Backend	Parity	200×30 (ms)
Python · pls4all
pls4all.python	✓ bind	1.99 ms
pls4all.sklearn	✓ 3e-15	1.41 ms🏆
R · pls4all
pls4all.R	✓ bind	4.09 ms
pls4all.R.formula	✓ bind	5.05 ms
pls4all.R.mdatools	✓ bind	4.73 ms
pls4all.R.pls	✓ bind	4.83 ms

---

See also: benchmark overview · methods index · interactive dashboard