rosario (version 0.1.0)

get_null_model: Null-model test via ROSARIO algorithm randomization

Description

Generates a null distribution of concurrent temporal niche overlap by repeatedly randomizing the input matrix with rosario_sample() and recomputing the mean pairwise overlap (see temp_overlap()).

Usage

get_null_model(mat, method, nsim = 1000, parallel = FALSE)

Value

A list with components:

observed_niche_overlap

Mean from all possible pairwise comparisons among biological identities for mat.

p_value

A one-sample t.test object comparing null means to the observed value (mu = observed).

null_niche_overlap

A tibble/data.frame of simulated mean overlaps (one per randomization).

Arguments

mat

Numeric matrix (rows = biological identities, columns = ordered time intervals).

method

Character string naming the overlap index to use: "pianka" or "czekanowski".

nsim

Integer number of randomizations to run (default 100).

parallel

Logical. If TRUE, randomizations are executed in parallel using furrr::future_map_dfr() with a multisession plan set internally.

Details

For "czekanowski", rows are rescaled to proportions internally to satisfy the index's assumptions. Randomization preserves each row's temporal autocorrelation by cyclically shifting (and optionally mirroring) its profile; see rosario_sample().

When parallel = TRUE, the function calls future::plan(multisession) internally so that randomizations are distributed across available local sessions. This means the function overrides any previously set future plan. If you need custom control over parallelization (e.g. cluster backends), run the non-parallel mode (parallel = FALSE) and handle parallelism externally.

See Also

temp_overlap(), rosario_sample(), temp_overlap_matrix()

Examples

get_null_model(ex1, method = "pianka", nsim = 10, parallel = FALSE)