Hidden logistic regression: state of the art

In 2003 Andreas Christmann and Peter J. Rousseeuw published a paper where they introduced what they called Hidden Logistic Regression , a model that was meant to help dealing with perfect prediction and outliers in logistic regression models − what is known as the Hauck-Donner phenomenon .

An R package was subsequently implemented, called hlr. As of today however, this package no longer exists. Does anyone know why? Was the method abandoned because it had flaws? Are there other methods known today to deal with this phenomenon?

Thanks in advance for any ideas.

PS: I posted this question on Stack Exchange where it was poorly received, I hope that this is a better place to ask, if not, my apologies.

It exists (its archived), and the reason was given at your link... Archived on 2015-06-19 as invalid maintainer address was not corrected despite reminders.
I.e. there was no person identified to shepard the package.

You can install the archived package if you choose

Thank you very much! This settles the R issue. Remains the question on what version of logistic regression is best to deal with separation issues - I found three (as discussed in this SE question) namely Hidden, Firth and Shen-Gao regressions, and I found a paper that claims that Shen-Gao is the best, but the paper is from 2012 and was only cited twice since then so I'd be a bit suspicious at first glance.

This topic was automatically closed 21 days after the last reply. New replies are no longer allowed.

If you have a query related to it or one of the replies, start a new topic and refer back with a link.