df_fevd: Fit EVD Parameters to Rows of a Data Frame of Annual Maxima

Description

Fits an extreme value distribution (EVD) to each row of a data.frame, where each row represents a time series of annual maxima. Optionally includes non-stationary covariates in the location parameter. Parallel processing is used to improve efficiency.

Usage

df_fevd(
  varbIN,
  cores = 1,
  ntries = 1,
  evd_mod_str = "fgumbel",
  nsloc = NULL,
  seed = NULL
)

Value

A data.frame with one row per input time series, containing the fitted EVD parameters returned by evd_params().

Arguments

varbIN: A numeric data.frame where each row is a vector of annual maxima values (e.g., one row per location or site).
cores: An integer specifying the number of parallel cores to use for computation.
ntries: The integer number of tries to fit the evd
evd_mod_str: A character string specifying the EVD fitting function to use. Must be one of "fgumbel", "fgumbelx", or "fgev".
nsloc: Optional data.frame of covariates for non-stationary location modelling. Must have the same number of rows as varbIN.
seed: Optional integer for reproducibility. If supplied, a random seed will be set before model fitting.

Details

Each row of varbIN is passed to evd_params(), optionally with a corresponding row of covariates from nsloc. Covariates are internally standardised if provided. The function constructs a blank parameter structure (empty_evd_params) to ensure consistent output structure, even if fitting fails.

Parallel computation is performed using parLapply.

Examples

Run this code

set.seed(42)
df <- as.data.frame(matrix(evd::rgev(1000), nrow = 100))
results <- df_fevd(df, cores = 1, evd_mod_str = "fgumbel")