In antononcube/ERTMon-R: Event Records Transformations Monad

library(dplyr)
library(Matrix)
library(ERTMon)

Introduction

This notebook demonstrates the application of formula specification data frame onto a features matrix.

The exposition has to be improved with easier to plot example results.

The feature sub-matrices

Run the notebook "Event-records-transformations-main-use-case.Rmd" first. (Better with the parameter categoricalMatricesQ set to FALSE.)

Here we take the feature matrix (sparse) sub-matrices from data transformation object (dtObj):

fakeMon <-
  ERTMonUnit() %>% 
  ERTMonReadDataFromDirectory( "../data/FakeData/" ) %>% 
  ERTMonProcessEventRecords(echoStepsQ = FALSE)

smats <- fakeMon %>% ERTMonTakeContingencyMatrices()
names(smats)

Specification

Here we make a fake formula specification for the fake medical data:

formulaSpecDF <- 
  data.frame( 
    "FormulaID" = c(1),
    "TermID" = c(1,1,1,2,2), 
    "TermCoefficient" = c( 0.5, 0.5, 0.5, 1.2, 1.2),
    "FeatureName" = c("HR.Mean", "NBP.M.Mean", "RR.Mean",  "NBP.M.Range", "NBP.M.Mean"),  
    "ReduceFunction" = c( "+", "+", "+", "*", "*"),
    "Coefficient" = c(1, 1, 1, 1, 1.6), 
    "Exponent" = c(1, 1, 2, 1, 1),
    "RatioPart" = c( "Numerator", "Denominator", "Denominator", "Numerator", "Denominator"),
    stringsAsFactors = FALSE)
formulaSpecDF

Each term in the formula specification has its own reduce function: "+" (summation) or "*" (multiplication).

Here is the interpretation (per term):

$$formulaMat = \frac{reduceFunc[ { Coefficient[i] * smats[ FeatureName[i] ] ^ {Exponent[i]} : i \in NumeratorRows } ]}{reduceFunc[ { Coefficient[i] * smats[ FeatureName[i] ] ^ {Exponent[i]} : i \in DenominatorRows } ] }$$

Application

The application of the formula specification is done with function ApplyFormulaSpecification.

fakeMon <- fakeMon %>% ERTMonComputeFormula( formulaSpec = formulaSpecDF )
resMat <- fakeMon %>% ERTMonTakeValue()

image(resMat)

antononcube/ERTMon-R documentation built on Oct. 14, 2021, 2:27 p.m.