random_frog_select — Random Frog

Group: Variable selector · Registry tolerance: 1e-06

Description

Random Frog selection (§18 Phase 5g)

From the pls4all.sklearn.RandomFrogSelector docstring:

Random Frog feature selection (Li 2012).

Registry note — Python auswahl.RandomFrog (LSX-UniWue; Li 2012). Same algorithm as libPLS randomfrog_pls. Default _random_frog_select_pls4all path mirrors the same auswahl call with random_state=seed, giving bit-exact mask parity. The C++ splitmix64 kernel is opt-in via legacy=True.

Parameters

Name	Type	Default	Notes
top_k	int	None	Number of features to retain.
n_components	int	2	Number of latent components extracted (k).
n_iterations	int	100	Number of selection iterations or Monte-Carlo passes.
initial_size	int	20	Starting subset size for the random-frog chain.
min_size	`int	None`	None
max_size	`int	None`	None
n_folds	int	3	Number of cross-validation folds used inside the selector.
seed	int	0	Random seed for reproducible sampling/initialization.

Explanations

Bibliographic source

Li, H., Xu, Q. & Liang, Y. (2012). Random frog: an efficient reversible jump Markov chain Monte Carlo-like approach for variable selection. Analytica Chimica Acta 740, 20–26.

Mathematical principle

Random Frog runs a random walk over feature subsets: at each step it proposes a transition to a neighbouring subset (add / remove / swap a feature) and accepts the transition with a Metropolis-style probability based on the improvement in CV-RMSE. Features that appear frequently in the visited subsets are deemed important.

Output: the inclusion frequency vector — fraction of iterations in which each feature was selected. Sort by frequency and take the top-\(k\) for the final subset.

Random Frog is sample-efficient compared to GA-PLS (no population of full subsets to maintain) but slower to mix on very high-dimensional data. Recommended on spectra of moderate size (a few hundred wavelengths).

Implementation

n4m_random_frog_select.

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_random_frog_select_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 random_frog_select_fit
with pls4all.Context() as ctx, pls4all.Config() as cfg:
    res = random_frog_select_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 RandomFrogSelector
mdl = RandomFrogSelector(top_k, n_components=2, n_iterations=100, initial_size=20, min_size=None, max_size=None, n_folds=3, seed=0)
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("random_frog_select", X, y,
                      n_components = 4L, params = list(n_iterations = 10L, initial_size = 10L, min_size = 5L, max_size = 20L, top_k = 10L, seed = 11L))
# res is a named list with MethodResult arrays/scalars.
# selected_indices / top_k_intervals are 1-based.

MATLAB · pls4all (MEX)

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

MATLAB · pls4all (classdef)

No idiomatic classdef wrapper — invoke pls4all.fit("random_frog_select", X, y, …) directly from the unified MEX factory.

Registry parity references 📐

• 📐 ref.python_auswahl (python · python) — auswahl 0.9.0 · strict (rmse_rel ≤ 1e-06) — Python auswahl.RandomFrog (LSX-UniWue; Li 2012). Same algorithm as libPLS randomfrog_pls with pinned random_state for bit-exact mask parity.

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 🏆.

Reference gate: strict — numeric equivalence (rmse_rel_tol ≤ 1e-06).

Rows tagged with 📐 are the canonical parity references for this method (declared in parity_timing.registry). C++ and external rows show reference parity; pls4all language bindings show binding parity against the C++ backend. Hover the icon for role and tolerance band.

1 thread

Backend	Parity	200×40 (ms)
Python · pls4all
pls4all.sklearn	✓ J 0.18	2.19 ms🏆
R · pls4all
pls4all.R	✓ J 0.18	4.98 ms
pls4all.R.formula	✓ J 0.18	5.67 ms
pls4all.R.mdatools	✓ J 0.18	5.96 ms
pls4all.R.pls	✓ J 0.18	5.94 ms

3 threads

Backend	Parity	200×40 (ms)
Python · pls4all
pls4all.sklearn	✓ J 0.18	2.19 ms🏆
R · pls4all
pls4all.R	✓ J 0.18	4.83 ms
pls4all.R.formula	✓ J 0.18	5.64 ms
pls4all.R.mdatools	✓ J 0.18	5.77 ms
pls4all.R.pls	✓ J 0.18	5.73 ms

10 threads

Backend	Parity	200×40 (ms)
Python · pls4all
pls4all.sklearn	✓ J 0.18	2.19 ms🏆
R · pls4all
pls4all.R	✓ J 0.18	5.15 ms
pls4all.R.formula	✓ J 0.18	5.73 ms
pls4all.R.mdatools	✓ J 0.18	5.76 ms
pls4all.R.pls	✓ J 0.18	5.71 ms

---

See also: benchmark overview · methods index · interactive dashboard