search for: epow

Displaying 2 results from an estimated 2 matches for "epow".

Did you mean: epo
2010 Nov 03
3
optim works on command-line but not inside a function
...University MSc student. ------ PS: the other functions I am using are: #IRT.p is the function that represents the probability #of a positive outcome of an item with difficulty b, #discriminativity a, in combination with a student with #competence theta. IRT.p <- function(theta, a, b){ epow <- exp(-a*(theta-b)) result <- 1/(1+epow) result } # = IRT.p^-1 ; for usage in the loglikelihood IRT.oneOverP <- function(theta, a, b){ epow <- exp(-a*(theta-b)) result <- (1+epow) result } # = (1-IRT.p)^-1 ; for usage in the loglikelihood IRT.oneOverPneg &l...
2011 Oct 05
4
SPlus to R
...-d)*nc))/nc + ncl<-length(pc) + ne.old<-ne + ne.old1<-ne.old + ### sweeping forward + for(i in 1:ncl){ + cmin<-cc[1] + cmax<-cc[2] + ### fixed cci<-cmax bug + cci <-1 + lhood<-dbinom((i:ncl)-1,nc,pc[i]) + ne[i:ncl]<-(1+(cci-1)*(lhood/lhood[1])) * ne.old1[i:ncl] + Ep0 <-Epower(nc, d, ne, pc, alpha) + while(abs(Ep0[i]-power)>tol1){ + if(Ep0[i]<power) cmin<-cci + else cmax<-cci + cci<-(cmax+cmin)/2 + ne[i:ncl]<-(1+(cci-1)*(lhood/lhood[1])) * ne.old1[i:ncl] + Ep0<-Epower(nc, d, ne, pc, alpha) + } +  ne.old1<-ne + } + ne1<-ne + ### sweeping backw...