mbinreg {lazy.stat}

R Documentation

Monotone Binary Regression

Description

Monotone Binary Regression

Usage

mbinreg(f, n, x = 1:length(f), method = 4, print = 0, plot = 0)

Arguments

f	vector of the frequency counts
n	vector of the # of trials
x	vector of the ordinal regressor variable
method	= 3 | 4 or "primary" | "secondary" or "continuous" | "discrete"
print	= 1 to print the result
plot	= 1 to plot the result

Details

This function finds phat which maximizes
log likelihood = f*log(phat) + (n-f)*log(1-phat)
subject to order(x) == order(phat)

The PAVA algorithm is ued.

The following code should produce the same result:
library(isotone)
res <- gpava( z=x, y=f/n, weights=n, ties=tie handling method )$x

References

de Leeuv, J., Hornik, K., and Mair, P. (2009) Isotone Optimization in R: Pool-Adjacent-Violators Algorithm (PAVA) and Active Set Methods. Journal of Statistical Software. Volume 32, Issue 5.

Examples

set.seed(1701)
nobs <- 15
# x <- 1:nobs
x <- floor(10*runif(nobs))
f <- round(10*runif(nobs)+1.5*x + 0.5)
f[f<0] <- 1
n <- rep( max(f)+1, nobs )
n <- n-sample(-2:2, nobs, replace=1)
f[f>=n]=n[f>=n]-1
f[f<0]=1

p0=f/n

res <- mbinreg( f, n, x, print=1, plot=1 )
res <- mbinreg( f, n, x, print=1, plot=1, method="p" )

---