knnAUC: knnAUC (k-nearest neighbors AUC test)

In liyistat/knnAUC: k-nearest neighbors AUC test

Description

knnAUC : k-nearest neighbors AUC test. Testing dependence/correlation of two variables is one of the fundamental tasks in statistics. In this work, we proposed a new way of testing nonlinear dependence between one continuous variable (X) and one binary variable (Y). We addressed this research question by using R package 'knnAUC'. In the knnAUC framework, we first calculated the AUC estimator based on a k-nearest neighbors classifier, and then evaluate the significance of the AUC based statistic (Alternative hypothesis: AUC > 0.5, that is to say, X has prediction power for Y).

Usage

knnAUC(x, y, ratio = 0.46, k = 100)

Arguments

x	A vector containing values of a continuous variable (X).
y	A vector containing values of a binary (0 or 1) discrete variable (Y).
ratio	A number, the training sample size ratio.
kmax	A positive integer, we'll automatically find the best parameter for knn between 1 and number kmax.

Value

result A list of AUC estimator and pvalue of knnAUC test.

Author(s)

Yi Li, [e-mail]

References

Yi Li, Xiaoyu Liu, et al. knnAUC : k-nearest neighbors AUC test.BMC Bioinformatics. 2018.

Examples

#Example1
#Generate random variable X and random binary variable Y
 set.seed(1234)
 x<-rnorm(100, mean = 0, sd = 1)
 y<-rbinom(100, size = 1, prob = 0.5)

#Test whether X and Y are independent or not using knnAUC method
#Calculate the average AUC estimator and pvalue of knnAUC method
#Set the parameters:ratio=0.5, k=50; And show the running time of knnAUC
 start.time <- Sys.time()
 library(knnAUC)
 knnAUC(x,y,ratio=0.5,k=50)
 end.time <- Sys.time()
 end.time - start.time


#Example2
#Generate random variables X and Y from logistic regression simulation
 set.seed(1234)
 N <- 100
 intercept <- 1
 beta <- 1
 x <- rnorm(N,mean=0,sd=1)
 linpred <- intercept + (x*beta)
 prob <- exp(linpred)/(1 + exp(linpred))
 runis <- runif(N,0,1)
 y <- ifelse(runis < prob,1,0)

#Test whether X and Y are independent or not using knnAUC method
#Calculate the average AUC estimator and pvalue of knnAUC method
#Set the parameters:ratio=0.46, k=100; And show the running time of knnAUC
 library(knnAUC)
 start.time <- Sys.time()
 knnAUC(x,y,ratio=0.46,k=100)
 end.time <- Sys.time()
 end.time - start.time

liyistat/knnAUC documentation built on May 12, 2019, 10:51 a.m.