In Rkabacoff/qacStats: Statistical functions for QAC courses

The plotROCS function^[Some code for the generation of the plot and relevant statistics as queried by the hidden helper function getProbs have been used from "Enhanced ROC Curve" coding example curated by Professor Robert Kabacoff as part of QAC385: Advanced R. Citation: Original source unknown. Retrieved by Kabacoff, Robert. "code - enhanced ROC curve.R." Function. Wesleyan University. Middletown, CT. October 2018. Accessed 4 December, 2018] is a more detailed alternative to plotting logistic regression model ROC curves than using the ROC package. The function inputs one or more data models, and calculates a detailed ROC curve with user specified options. These options include line and plot color manipulation, generation of the optimal point on plot, and cutoff value displays. A table of important values for each model is printed to the console upon running; the values include the optimal cutoff point, model accuracy, sensitivity, specificity, and area under the curve. Plot dimensions depend on the "Plots" window size where graph is generated.

Note: The data sets for each model must be unique. Using models generated from the same data set (ex, train) will result in an error.

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)

Set Up and a Few Examples

There are two preprogrammed model examples available to this function; one using the regclass data set PIMA^[Retrieved from Course Moodle Page for QAC385: Advanced R. Kabacoff, Robert. "code - Logistic Regression 3.R." Code sample. Wesleyan University. Middletown, CT. October 2018. Accessed 4 December, 2018], and another utilizing Wisconsin Breast Cancer data^[Retrieved from Course Moodle Page for QAC385: Advanced R. Kabacoff, Robert. "code - Logistic Regression 2.R." Code sample. Wesleyan University. Middletown, CT. October 2018. Accessed 4 December, 2018].

library(plotROCS)
#PIMA
fit.glm <-model1()
#Breast Cancer
fit<-model2()

Usage

Simply input your models into the function. An example with all of the presets:

plotROCS(fit.glm, fit)

An example changing the line and optimal point colors:

plotROCS(fit.glm, fit, colorsList = c("purple","pink"), colorPoints = "green")

An example changing the cutoff values, excluding the optimal points:

plotROCS(fit.glm, colorsList = "cornflowerblue", opCP = F, cutoffSpec = seq(0,1,by=.25))

Rkabacoff/qacStats documentation built on Jan. 17, 2024, 9:25 p.m.