Reduce time-series to few Principal Components

This function is under active development. It is meant to reduce the entire time-series of normalized and baseline-corrected pupillary data in just a few scores obtained by Principal Component Analysis. PCA is an effective way to reduce data dimensionality as to have more manageable dependent variables, which may additionally help having more precise estimates (or fingerprints) of pupil signal and the underlying cognitive processes.

Usage

reduce_PCA(
  data,
  dv,
  time,
  id,
  trial,
  Ncomp = NULL,
  center = FALSE,
  scale = FALSE,
  add
)

Arguments

data: A data.frame containing all the necessary variables.
dv: A string indicating the name of the dependent variable.
time: A string indicating the name of the time variable.
id: A string indicating the name of the id (participant) variable.
trial: A string indicating the name of the trial variable.
Ncomp: Number of components to retain. The default (NULL) automatically retains 95% of the explained variance. If Ncomp== "all" returns all the components. If Ncomp <1 this is interpreted as if the user wishes to retain a given proportion of variance (e.g. 0.6).
center: Whether variables, i.e. pupil size for each timepoint, should be scaled beforehand. Defaults to FALSE assuming that measures are already normalized (with z-scores) and baseline-corrected.
scale: Whether variables, i.e. pupil size for each timepoint, should be scaled beforehand. Defaults to FALSE assuming that measures are already normalized (with z-scores) and baseline-corrected.
add: String(s) indicating which variables names, if any, should be appendend to the scores dataframe.

Value

A list including the processed data, the scores and loadings dataframes, and the PCA object useful for prediction of new data.