R/curvetest.R

In curvetest: The package will formally test two curves represented by discrete data sets to be statistically equal or not when the errors of the two curves were assumed either equal or not using the tube formula to calculate the tail probabilities.

curvetest <-
function (formula,   data1=NULL, data2=NULL,  
   equal.var = TRUE, alpha =  0.5, bw=NULL, plotit = TRUE, conf.level = 0.05,  
    kernel = c("Triangle",  "Gaussian", "Trio","Uniform", "Triweight",  "Epanechnikov", 
    "Quartic"), nn=100, myx = NULL, bcorrect="simple",...) {
   if (missing(kernel)) kernel = "Quartic"
   if(length(kernel)>1) kernel = "Quartic"
   types<-c("Gaussian", "Trio",  "Uniform", "Triweight", "Triangle", "Epanechnikov", "Quartic")
   pm = pmatch(kernel, types)
   ww<-get.weight.function(types[pm]) 
   data1 <-model.frame(formula=formula, data=data1); 
   if(!is.null(data2)) data2 <-model.frame(formula=formula, data=data2); 
   if(missing(alpha)) alpha=0.5 
   if(is.null(data2)) myx<-seq(min(data1[,2]), max(data1[,2]), length=nn) else 
   if(is.null(myx)) myx<-seq(min(data1[,2], data2[,2]), max(data1[,2], data2[,2]), length=nn)    
   if(!is.null(alpha)){
     if(is.numeric(alpha)) alpha1= alpha else  
     if(alpha=="optimal")  alpha1= as.numeric(getoptimalalpha(formula=formula, data1)) 
   }else  alpha1=NULL
   f1<-curvefit(formula=formula, data=data1, kernel=kernel, alpha=alpha1,bw=bw,bcorrect=bcorrect, myx=myx)   
   f2<-NULL;
   if(!is.null(data2)) {
     if(!is.null(alpha)) {
     if(is.numeric(alpha)) alpha2= alpha else  
     if(alpha=="optimal")  alpha2= as.numeric(getoptimalalpha(formula=formula, data2)) 
     }else  alpha2=NULL
   f2<-curvefit(formula=formula, data=data2, kernel=kernel,  alpha=alpha2, bw=bw,myx=myx, bcorrect=bcorrect)
     out<-curvetest.raw(fits1=f1, fits2=f2,equal.var =equal.var,   
       conf.level = conf.level, plotit = plotit)
   }
   if(plotit) plot(x=f1, col=2,...)
   if(plotit&!is.null(data2)) plot(x=f2,add=T, col=3, ...)
   return(out)
}