Fit the truncated path-signature single-sample discriminator

Maps the TRAINING matrix to the per-sample robust CLR over the frozen feature universe (colnames(X_train)), computes each training specimen's closed-form truncated path signature (time-augmented 2D path, level L, pure base R, ROW-BY-ROW), freezes a per-coordinate signature-coefficient standardization, and fits a FROZEN ridge-logistic head over the standardized coefficients. If hp$lambda is NULL the ridge penalty is selected by TRAINING-ONLY deterministic class-stratified k-fold CV (no RNG). The fitted model holds NO python dependency and NO external pointer.

Usage

fit_sig_path(X_train, y_train, meta_train = NULL, hp = list())

Arguments

X_train

Numeric matrix (samples \(\times\) features) of non-negative abundances with unique, non-empty feature names (colnames). At least 2 features are required (a path needs at least one segment).

y_train

Numeric / integer 0/1 labels (1 = case), length nrow(X_train), with at least one case and one control.

meta_train

Optional per-sample metadata. Accepted for interface uniformity and otherwise ignored.

hp

Optional list of hyperparameters. Allowed fields: L (signature truncation level, one of 1L, 2L, 3L, 4L; default 3L; signature dimension \(D = \sum_{k=1}^{L} 2^k\), so 14 at L = 3), lambda (ridge penalty for the logistic head; NULL (default) selects it by training-only deterministic k-fold CV, otherwise a non-negative finite number), min_features (feature-overlap floor at scoring, positive integer; default 3L), and seed (integer; default 42L). NO device hp — the method is pure base R.

Value

Object of class sig_path_model: a list with feature_universe, L, D (signature dimension), coef_mean, coef_sd (frozen standardization), intercept, weights (frozen linear head), lambda (resolved ridge penalty), seed, and hp. No python dependency, no external pointer.

References

Chevyrev I, Kormilitzin A. (2016) A primer on the signature method in machine learning. arXiv:1603.03788.

Examples

if (FALSE) { # \dontrun{
set.seed(1)
n <- 80; p <- 16; k <- 6
L <- matrix(stats::rnorm(n * p, 4, 0.5), nrow = n,
            dimnames = list(NULL, paste0("miR-", seq_len(p))))
y <- rep(c(0, 1), each = n / 2)
L[y == 1, seq_len(k)] <- L[y == 1, seq_len(k)] + 1.2
X <- exp(L)
model <- fit_sig_path(X, y)
score_sig_path(model, X[1, , drop = FALSE])
} # }