extended_lnlp: Perform regularized S-map using a generalized function
In ong8181/macam: A collection of miscellaneous R functions for ecology

extended_lnlp

R Documentation

Perform regularized S-map using a generalized function

Description

extended_lnlp performs the regularized S-map introduced in Censi et al. (2019) Methods in Ecology and Evolution. Multivariate S-map is also supported.

Usage

extended_lnlp(
  block_time,
  lib = c(1, NROW(block_time)),
  pred = lib,
  tp = 1,
  target_column = 1,
  lib_column = 1:NCOL(block_time),
  num_neighbors = NCOL(block_time) + 1,
  theta = 0,
  dist_w = NULL,
  regularized = FALSE,
  lambda = NULL,
  alpha = 0,
  glmnet_parallel = FALSE,
  random_seed = NULL,
  save_smap_coefficients = FALSE
)

Arguments

`block_time`	Dataframe or matrix. Original time series.
`lib`	Numeric vector. Library indices.
`pred`	Numeric vector. Prediction indices.
`tp`	Forecasting time ahead.
`target_column`	Numeric. Indicates target column
`lib_column`	Numeric. Indicates library column
`num_neighbors`	Numeric. The number of nearest neighbors.
`theta`	Numeric. Weighing function for S-map.
`dist_w`	Matrix. Distance matrix used to calculate weights for S-map. Implemented for MDR S-map (Chang et al. 2021) Ecology Letters. If `NULL`, then weights are calculated based on Euclidean distance.
`regularized`	Logical If `TRUE`, regularized S-map will be performed. If `FALSE`, the normal S-map will be performed. Please use `rEDM::s_map` function.
`lambda`	Numeric. Specify the strength of penalty in the regularization.
`alpha`	Numeric. `alpha = 0` is the ridge regression, `alpha = 1` is the lasso regression, and `⁠0 < alpha < 1⁠` is an elastic net.
`glmnet_parallel`	Logical. If TRUE, the computation will be parallel (currently, experimental).
`random_seed`	Numeric. Random seed.
`save_smap_coefficients`	Logical. If `TRUE`, S-map coefficients will be saved.

Details

Cenci, S, Sugihara, G, Saavedra, S. Regularized S-map for inference and forecasting with noisy ecological time series. Methods Ecol Evol. 2019; 10: 650– 660. https://doi.org/10.1111/2041-210X.13150