ROC Surface Analysis Under the Three-Class Problems.
ROCsurf
The objective of this project is to assess the Receiver Operating Characteristic (ROC) surfaces for Gamma, Weibull, and Logistic distributions. Additionally, it supports performance evaluations linked to these ROC surfaces in the context of three-class problems.
Installation
You can install the development version of ROCsurf via the following code:
# install.packages("devtools")
devtools::install_github("ErtanSU/ROCsurf")
Example
This is a basic example which shows you how to solve a common problem:
library(ROCsurf)
dG(c(1,2,3,4,5,200,1000),alpha=6,beta=.8)
#> [1] 9.107742e-03 8.350118e-02 1.816690e-01 2.193342e-01 1.917734e-01
#> [6] 2.715241e-99 0.000000e+00
dW(c(1,2,3,4,5,200,10000),alpha=1,beta=2)
#> [1] 3.032653e-01 1.839397e-01 1.115651e-01 6.766764e-02 4.104250e-02
#> [6] 1.860038e-44 0.000000e+00
dL(c(1,2,3,4,5,200),alpha=1,beta=.1)
#> [1] 2.500000e+00 4.539581e-04 2.061154e-08 9.357623e-13 4.248354e-17
#> [6] 0.000000e+00
library(ROCsurf)
pG(c(.5,1,2,3,4,25),alpha=6,beta=.8)
#> [1] 4.859954e-05 1.838085e-03 4.202104e-02 1.771172e-01 3.840393e-01
#> [6] 1.000000e+00
pW(c(.5,1,2,3,4,100),alpha=1,beta=2)
#> [1] 0.2211992 0.3934693 0.6321206 0.7768698 0.8646647 1.0000000
pL(c(.5,1,2,100),alpha=1,beta=.1)
#> [1] 0.006692851 0.500000000 0.999954602 1.000000000
library(ROCsurf)
qG(c(.9971,0.5,0.3),alpha=6,beta=.8)
#> [1] 11.956348 4.536129 3.613711
qW(c(.9971,0.5,0.3),alpha=1,beta=2)
#> [1] 11.6860891 1.3862944 0.7133499
qL(c(.9971,0.5,0.3),alpha=1,beta=.1)
#> [1] 1.5840140 1.0000000 0.9152702
library(ROCsurf)
rG(10,alpha=6,beta=.8)
#> [1] 6.342921 5.613297 4.344726 3.548187 4.781843 2.190594 1.504722 8.491058
#> [9] 1.787702 2.788674
rW(10,alpha=1,beta=2)
#> [1] 0.3372294 3.5360746 0.9712284 2.0038895 1.2751727 1.6049933 0.2083131
#> [8] 0.0378331 1.0150278 3.3648067
rL(10,alpha=1,beta=.1)
#> [1] 0.6838594 1.4337235 0.9796567 1.0586671 1.0282047 1.1179200 1.0458565
#> [8] 1.1385654 0.9166187 0.8226117
library(ROCsurf)
x<-rW(100, 2, 1)
y <- rG(100, 2, 2)
z <- rW(100, 6, 9)
r.tc_vus(x=x,y=y,z=z,
init_param=c(alpha1=2,beta1=1,alpha2=2,beta2=2,
alpha3=6,beta3=9),
model=c("WGW"), method=c("MLE"))
#> [1] 0.810257
library(ROCsurf)
x<- rW(100, 2, 1)
y <- rG(100, 2, 2)
z <- rW(100, 6, 9)
r.tc_index(x=x,y=y,z=z,
init_param=c(alpha1=2,beta1=1,alpha2=2,
beta2=2,alpha3=6,beta3=9),
init_index=c(median(x),median(y)),
model=c("WGW"),
method=c("MLE"))
#> c₁ c₂ TPF₁ TPF₂ TPF₃
#> J 1.765644 6.034736 0.9533556 0.6212299 0.9249205
#> PM 1.486714 6.668990 0.8760312 0.7143641 0.8606857
#> MV 1.658977 6.301341 0.9305363 0.6601923 0.9016357
#> NI 1.671323 6.282483 0.9335618 0.6566622 0.9034494
#> M 1.535361 6.562809 0.8939014 0.6994002 0.8736471
library(ROCsurf)
x<- rW(100, 2, 1)
y <- rG(100, 2, 2)
z <- rW(100, 6, 9)
r.tc_graph(x=x,y=y,z=z,
init_param=c(alpha1=2,beta1=1,alpha2=2,
beta2=2,alpha3=6,beta3=9),
empirical=FALSE,model=c("WGW"),
method=c("MLE"))
Corresponding Author
Department of Statistics, Faculty of Science, Selcuk University, 42250, Konya, Turkey
Email:https://www.researchgate.net/profile/Ertan-Akgenc
References
Akgenç, E., and Kuş, C., 2023, Statistical Inference for ROC Surface Analysis Under the Three-Class Problems, 7th International Congress of Researchers, Statisticians and Young Statisticians (IRSYSC-2023).
B. R. Mosier and L. E. Bantis., 2021, Estimation and construction of confidence intervals for biomarker cutoff-points under the shortest euclidean distance from the roc surface to the perfection corner, Statistics in medicine, 40(20):4522–4539.
G. Jameson., 2016, The incomplete gamma functions., The Mathematical Gazette, 100(548):298–306.
T. Dong., 2014, Selected Topics on Statistical Methods for Three and Multiple Class Diagnostic Studies, State University of New York at Buffalo.
J. Luo and C. Xiong., 2013, Youden index and associated cut-points for three ordinal diagnostic groups, Communications in Statistics-Simulation and Computation, 42(6):1213–1234.
F. Edition, A. Papoulis, and S. U. Pillai., 2002, Pobability, random variables, and stochastic processes, McGraw-Hill Europe: New York, NY, USA.
A. J. Hallinan Jr., 1993, A review of the weibull distribution, Journal of Quality Technology, 25(2):85–93.
N. Balakrishnan., 1991, Handbook of the logistic distribution, CRC Press.