R/default.knots.r

In AdaptFitOS: Adaptive Semiparametric Additive Regression with Simultaneous Confidence Bands and Specification Tests

# Computes default knots for a given x vector.


default.knots <- function(x,num.knots,knotchoice="quantiles"){
   # Delete repeated values from x
   x <- unique(x)

   # Work out the default number of knots
   if (missing(num.knots))   {
      n <- length(x)
      d <- max(4,floor(n/35))
      num.knots <- floor(n/d - 1)
   }

   nx <- names(x)
   x <- as.vector(x)
   nax <- is.na(x)
   if(nas <- any(nax)) x <- x[!nax]

   if(knotchoice=="equidistant")
     knots = seq(min(x), max(x), length=num.knots)
   else if(knotchoice=="quantiles")
     knots <- quantile(x,seq(0,1,length=num.knots+2)[-c(1,num.knots+2)], names=F)
   else
     stop("Unknown knot choice.\n")
   
   names(knots) <- NULL
 
   return(knots)
}