fitCorModel: Fitting a correlation model by least-squares

In martindurocher/floodRFA: A R-package for regional frequency of floods

Description

Returns the parameter of the correlation model

Usage

fitCorModel(h, ...)

## S3 method for class 'numeric'
fitCorModel(h, x, type = "exp", w = NULL, nugget = 0,
  smooth = 1, hPred = NULL, startp = NULL, bound = NULL, ...)

## S3 method for class 'matrix'
fitCorModel(x, ...)

## S3 method for class 'formula'
fitCorModel(form, x, ...)

Arguments

h	Vector of distance
...	Other parameters pass to the funtion optim.
x	Vector of correlation coefficients or Nx2 matrix of the distance and correlation coefficients.
type	Correlation model. See corModel
w	Vector of weights.
nugget	The nugget parameter. If NULL the parameter is optimized.
smooth	The smooth parameter for three parameters model.
hPred	A vector of distance for prediction (optional)
startp	The starting point for optimization. May be used when the nugget effect is estimated.
form	Formula indication the correlation coefficient and the distance. For instance x~h

Examples

tt = 1:500
rr = corModel(c(200,.1), tt, type = 'exp') +
              rnorm(length(tt),0,.2)

df <- data.frame(rho = rr, dist = tt)

fitCorModel(tt, rr, type = 'sph')

sol <- fitCorModel(rho ~ dist, df, nugget = NULL,
                   type = 'exp', hPred = tt, startp = c(200,.1))

plot(tt,rr, ylab = 'rho', xlab = 'distance')
lines(sol$pred$dist, sol$pred$rho, col = 2, lwd = 2)

martindurocher/floodRFA documentation built on June 5, 2019, 8:44 p.m.