[R] bugs and misfeatures in polr(MASS).... fixed!

ahs ah.sjostrom at gmail.com
Thu Oct 18 14:05:19 CEST 2012

I am trying to use this fix for the convergence problem in polr, but I don't
seem to get the change of code right. I redefine the function polr by the

tjb wrote
>     if(missing(start)) {
>       # try something that should always work -tjb
>       u <- as.integer(table(y))
>       u <- (cumsum(u)/sum(u))[1:q]
>       zetas <-
>          switch(method,
>                 "logistic"= qlogis(u),
>                 "probit"=   qnorm(u),
>                 "cauchit"=  qcauchy(u),
>                 "cloglog"=  -log(-log(u)) )
>       s0 <- c(rep(0,pc),zetas[1],log(diff(zetas)))

but - which part should I erase? The whole old if(missing(start)) {...} or
something else or nothing? I have tried all ways I could think of.

I also can't see where the new object s0 is used in the old code?

Any input on my problem would be very much appreciated!

View this message in context: http://r.789695.n4.nabble.com/bugs-and-misfeatures-in-polr-MASS-fixed-tp3024677p4646600.html
Sent from the R help mailing list archive at Nabble.com.

More information about the R-help mailing list