Understanding the getScore Function Input

Overview

This vignette describes the usage of the getScore parameter in the pandemonium app.

This is an input for a function that calculates scores and bins which will be plotted in the input panel and can be used as colouring in dimension reduction and tour views.

The pandemonium package provides two functions for scores or one can be created and passed. The two functions provided are:

• chi2score This will calculate chi-squared values from the exp reference point in space1. The scores returned are chi-squared values, the bins are sigma bins from the reference point and the interesting points are ‘bf’, the best fit and ‘sm’, the closet point to the origin.
• outsidescore This function takes a set of scores for each point and a name for the score from data outside the app. It is passed when loading the app. the scores returned are the scores passed to it, the bins are quartile bins of that score and no interesting points are returned.

Writing a getScore function

Input

The app will pass the following objects to getScore that may be used in calculations:

Input	Description
space1	Data frame of space 1 values
cov	Covariance matrix of space 1
covinv	Inverse covariance matrix of space 1
exp	reference point in space 1
space2	Data frame of space 2 values
space2.cov	Covariance matrix of space 2
space2.exp	reference point in space 2
k	number of clusters being made can be used to set number of bins

Output

The function should return a named list with some of the following entries to be used by the app.

Output	Description
score	This should be a vector of length = number of rows in space 1, the values should be numeric
scoreName	A name for the title of the score plot
bins	This should be a factor vector with length = number of rows in space 1
binName	A name for the title of the bin plot
interest	A vector of names of interesting points to display in the hover of htmlwidgets plots
is.interest	A logical vector of interesting points to display in ggplot plots

Example

A simple template for what a getScore function should look like is as follows:

myScore <- function(space1=rv$space1, cov=rv$cov1, covinv=rv$covInv1,
         exp=rv$exp, space2=rv$space2, space2.cov=rv$cov2, 
         space2.exp=rv$space2.exp, k=rv$kC){
  
  ## Calculations
  n      <- nrow(space1)
  scores <- #function to calculate scores
  bins   <- #function to bin scores
  
  ## Return
  ret <- list()
  ret$score       <- scores
  ret$scoreName   <- "scores"
  ret$bins        <- bins
  ret$binName     <- "bins"
  ret$interest    <- rep("",n)
  ret$is.interest <- which(ret$interest!="")
  ret
}

pandemonium(df, getScore = myScore)

For the function to take values from outside the app, it is best to write it as a closure and pass these values when calling the app. This has been done for outsidescore as seen below:

outsidescore <- function(scores,scoreName = NULL){
  function(space1, ...){
    ret <- list()
    n<- nrow(space1)
    ret$score <- scores
    ret$bins <- cut(scores, stats::quantile(scores,c(0,0.25,0.75,1))-c(1,0,0,0), labels=c("lower","inner","upper"))
    ret$interest <- rep("",n)
    ret$is.interest <- which(ret$interest!="")
    ret$scoreName <- as.character(scoreName)
    ret$binName   <- "quartile"
    ret
  }
}

pandemonium(df = Bikes$space1, space2 = Bikes$space2, 
                getScore = outsidescore(Bikes$other$res,"Residual"))