Predicting property auction outcomes using logistic regression
by
Kyungjin Byun
Bond University
Coauthors: Professor John S Croucher

This paper looks at the attributes of over 140 properties that sold at auction in the Ku-Ring-Gai region of Sydney, Australia during 2001. Since this is rapidly becoming the preferred method of sale by many vendors in a number of areas of the country, a study is made to find just what factors contribute mostly to a successful outcome on the day. Logistic regression is used to determine the most significant attributes of these properties as well as the probability that a property with these qualities will actually sell. The models used can be readily applied to property auctions in any region of the world.

Date received: October 11, 2002