smoothSurv (version 2.6)

find.c: Work Function for 'smoothSurvReg'

Description

Find mixture proportions that approximate given distribution by a G-spline mixture.

Usage

find.c(knots, sdspline, dist = "dnorm")

Value

Either the vector of G-spline 'c' coefficients or NULL

if there are problems to find them.

Arguments

knots

A vector of G-spline knots \(\mu\).

sdspline

Standard deviation \(\sigma_0\) of the basis G-spline.

dist

A character string specifying the function used to compute a density of the distribution you want to approximate.

Author

Arnošt Komárek

Details

The function finds the G-spline coefficients that approximates a density given by dist in such sense that the value of the G-spline is exactly equal to the value of that density in knots.

Examples

knots <- seq(-4, 4, 0.5)
sd0 <- 0.3
ccoef <- find.c(knots, sd0, dist = "dstlogis")

### We plot the approximation together with the truth
###
grid <- seq(-4, 4, 0.05)
truth <- dstlogis(grid)

### Following lines compute the values of the approximation
grid.big <- matrix(grid, nrow = length(grid), ncol = length(knots))
knots.big <- matrix(knots, nrow = length(grid), ncol = length(knots), byrow = TRUE)
normals <- dnorm(grid.big, mean = knots.big, sd = sd0)
approx <- normals %*% ccoef

### Plot it
plot(grid, approx, type = "l", xlab = "y", ylab = "f(y)", bty = "n")
lines(grid, truth, lty = 2)
legend(-4, 0.35, c("approx", "truth"), lty = 1:2, bty = "n")