ps_pca: ps_pca

In benmarwick/karon: Compositional Data Analysis of Archaeological Artefacts

ps_pca

Description

Compute and plot principal components after standardizing the data

Usage

ps_pca(
  doc = "ps_pca",
  data,
  ID = " ",
  GroupVar,
  Groups,
  AnalyticVars,
  ScreePlot = FALSE,
  BoxPlots = FALSE,
  pcPlot = TRUE,
  PlotPoints = TRUE,
  PlotEllipses = TRUE,
  PlotHull = FALSE,
  PlotMedians = FALSE,
  Ellipses = c(0.95, 0.99),
  PlotColors = TRUE,
  legendLoc = "topright",
  Colors = c("red", "black", "blue", "green", "purple"),
  Identify = FALSE,
  digits = 3,
  Seed = 11111,
  folder = " "
)

Arguments

doc	A string with documentation in the list returned, default is the function name
data	A matrix or data frame containing the data to be analyzed
ID	An optional name for an ID, default is " " if no ID
GroupVar	The name for variable defining grouping; a variable name is required
Groups	Character-valued defining the values of the group variable for which plots are to be done. Options are a vector of values; "All" (use all groups). One of these is required
AnalyticVars	A vector of names (character values) of analytic results
ScreePlot	Logical, if TRUE create a scree plot, default is FALSE
BoxPlots	Logical, if TRUE, create box plots of the first two components, default is FALSE
pcPlot	Logical, if TRUE (the default), create the plot of the first two components
PlotPoints	Logical, if TRUE (the default) and pcPlot=TRUE, plot the points for the first two components
PlotEllipses	Logical, if TRUE (the default), plot the confidence ellipse or ellipses for each group
PlotHull	Logical, if TRUE, plot the convex hull for each group, default is FALSE
PlotMedians	Logical, if TRUE, plot the symbol for each group at the median point for that group, default is FALSE
Ellipses	A value or vector of proportions for confidence ellipses; default is c(.95,.99) to produce 95% and 99% confidence ellipses
PlotColors	Logical. If TRUE, use list of colors in Colors for points; if F, plot points as black
legendLoc	Character, location of legend for a plot with points; default is "topright", alternatives are combinations of "top", "bottom", "right", "left"
Colors	A vector of color names; default is a vector with five names
Identify	Logical. If TRUE, the user can identify points of interest in plots; information on these points is saved to a file; default is FALSE
digits	The number of significant digits to return in objects in data frames, default is 3
Seed	If not NA, the seed for the random number generator used if missing data are imputed; default is 11111
folder	The path to the folder in which data frames will be saved; default is " "

Value

The function produces a plot of the first two principal components, the contents of which are defined by the arguments PlotPoints, PlotEllipses, PlotHull, and PlotMedians. A scree plot and box plots are produced if requested. The function returns a list with the following components:

• usage: A string with the contents of the argument doc, the date run, the version of R used

• dataUsed: The contents of the argument data restricted to the groups used

• dataNA: A data frame with observations containing a least one missing value for an analysis variable, NA if no missing values

• params: A list with the values of the arguments for grouping, logical parameters, Ellipses, and Colors

• analyticVars: A vector with the value of the argument AnalyticVars

• ellipse_pct: The value of the argument Ellipses

• variances: A data frame including the percent of variation explained by each principal component and the cumulative percent explained

• weights: A data frame with the principal component weights for each observation

• Predicted: A data frame with the predicted values for each principal component, plus the value of Groups and an integer GroupIndex (with values 1:number of Groups)

• DataPlusPredicted: A data frame with the data used to compute the principal components, plus GroupIndex (as defined above) and predicted values for each principal component

• dataCheck: If Identify=TRUE, a data frame with the observations in dataUsed identified as of interest

• location: The value of the parameter folder

Details

If Identify=TRUE, the user must interact with each plot (or pane, if there is more than one pane on a plot). To identify a point, place the cursor as close as possible to the point and left click; repeat if desired. To go to the next pane, right click and select "Stop" in base R; click on "Finish" in the plot pane in Rstudio.

Examples

data(ObsidianSources)
analyticVars<-c("Rb","Sr","Y","Zr","Nb")
save_pca <- ps_pca(data=ObsidianSources, ID="ID", GroupVar="Code",
Groups="All", AnalyticVars=analyticVars)

benmarwick/karon documentation built on July 29, 2023, 10:11 a.m.