Topic: Utilities

Topic Description:

Basic R utilities used in many packages and functions, such as date and string manipulations, statistical distributions, geometry of lines and distances. The R package date is required for the two data functions.



File: utilities/statistics.r

View File Source Download File No help file available

Function: llike.GaussModelSD

Function Description: llike.GaussModelSD

This is for model.xy. Take the function for the SD, its parameters, and both predicted and observed values of the dependent variable (pred,obs) to generate a likelihood. One of the parameters is passed as testparam, for use with metrop1step. The predicted value for each observation is included, and not calculated from the predicting function. MINIMUM_SD=.0001 MINIMUM_SD should be set in the program calling model.xy, to adjust it appropriately. If MINSD==0, then the sd can collapse to a miniscule number and drive the likelihood very high, preventing parameter searches from ever escaping the sd.

Function Arguments:

ArgumentDefault Value
testparam
allparam
whichtest
x
pred
obs
model
badsd
...

Function Source:

llike.GaussModelSD=function(testparam,allparam,whichtest,x,pred,obs,model,badsd,...)
{
extra=list(...)
if(is.null(extra$MINIMUM_SD)) MINIMUM_SD=0
else MINIMUM_SD=extra$MINIMUM_SD

param=arrangeParam.llike(testparam,allparam,whichtest)

sd=model(x=x,param=param)
if(length(which(sd<=MINIMUM_SD))>0) return(-Inf)
if(!is.null(badsd)) if(badsd(x,param)) return(-Inf)

llike=dnorm(obs,mean=pred,sd=sd,log=TRUE)
total=sum(llike)
if(is.na(total) | is.infinite(total) | is.null(total)) browser()

return(total)
}