A_missing: Calculate the sample loadings for samples not involved in...

In MatthiasPucher/staRdom: PARAFAC Analysis of EEMs from DOM

Calculate the sample loadings for samples not involved in model building

Description

Samples from an eemlist that were not used in the modelling process are added as entries in the A-modes. Values are calculated using fixed B and C modes in the PARAFAC algorithm. B and C modes can be provided via a previously calculated model or as matrices manually.

Usage

A_missing(
  eem_list,
  pfmodel = NULL,
  cores = parallel::detectCores(logical = FALSE),
  adj_samples = TRUE,
  components = NULL,
  const = NULL,
  control = NULL,
  ...
)

Arguments

eem_list	object of class eemlist with sample data
pfmodel	object of class parafac
cores	number of cores to use for parallel processing
adj_samples	logical, whether wavelengths in the samples are cut automatically to fit the present data in the model
components	optionally supply components to use manually, either as a variable of class parafac_components or as a list of variables of class parafac_components, if you do so,
const	optional constraints for model, just used, when components are supplied
control	optional constraint control parameters for model, just used, when components are supplied
...	additional arguments passed to eem_parafac

Details

This function can be used to calculate A modes (sample loadings) for samples that were previously excluded from the modelling process (e.g. outliers). The wavelengths in the samples and in the PARAFAC model must be identically. Surplus wavelengths in the samples can be adjusted automatically with the argument adj_samples = TRUE, surplus wavelengths in the PARAFAC model have to be mitigated by interpolating the missing wavelengths in the samples manually. If there is a mismatch in wavelengths, the function provides you with the necessary information. Another way to use it would be a recombination of components from different models and calculating the according sample loadings. Expecially the later application is experimental and results have to be seen critically! Nevertheless, I decided to supply this function to stimulate some experiments on that and would be interested in your findings and feedback.

Value

object of class parafac

Examples

data(eem_list)
data(pf_models)

A_missing(eem_list, pf4[[1]], cores = 2)

MatthiasPucher/staRdom documentation built on Nov. 9, 2024, 12:11 p.m.