Findbp: Search best hyperparameters for SA fitting (lambda and weight...

In ELAplus: Energy Landscape Analysis for Ecological Communities

Search best hyperparameters for SA fitting (lambda and weight decay) by cross-validation

Description

Repeating cross-validation over a grid of hyperparameters for simulated annealing-based fitting, evaluates prediction/fitting errors on held-out samples, and returns candidate best parameter/timepoint combinations.

Usage

Findbp(
  ocmat,
  enmat = NULL,
  cvmode = c("10fold", "5fold", "loo"),
  nrepeat = 5,
  we = c(0.001),
  maxlrs = c(0.005),
  totalit = 4000,
  lmd = c(5e-04, 0.001, 0.0025, 0.005, 0.0075),
  rep = 8,
  intv = 100,
  threads = 1,
  tol = 0.03,
  runadamW = TRUE,
  sparse = TRUE,
  fastfitting = TRUE,
  reporting = TRUE
)

Arguments

ocmat	Numeric matrix (0/1). occurrence matrix of the communities.
enmat	Optional environmental parameter matrix.
cvmode	Cross-validation mode. One of '"10fold"', '"5fold"', or '"loo"'. '"loo"' performs leave-one-out cross-validation. Default is '"10fold"'.
nrepeat	Number of repeated cross-validation runs for k-fold CV. Ignored when 'cvmode = "loo"'. Default is '10'.
we	Numeric vector. Candidate weight decay values.
maxlrs	Numeric vector. Maximum learning rate.
totalit	Integer. Total iterations for SA.
lmd	Numeric vector. Candidate lambda values.
rep	Integer. Number of replicates for runSA.
intv	Integer. Recording interval.
threads	Integer. Number of cores.
tol	Numeric. Error tolerance for finding the best hyperparameters (default: 0.03)
runadamW	Logical. Use AdamW.
sparse	Logical. Use sparsity penalty.
fastfitting	Logical. If TRUE, returns early iterations within the range specified by the parameter tol (default: 0.03) of best error.
reporting	Logical. Print progress.

Value

A list with:

best_candidates

A named list of best hyperparameter settings and scores.

allresults

All validation results for each grid point.

ELAplus documentation built on May 28, 2026, 5:09 p.m.