Department of Statistical Science
Duke University
presents:
Ed Iversen
iversen@stat.duke.edu
Duke Statistics
Duke University
"A Spatial and Temporal Markov Random Field Model with Application to Real Estate Indices"
Abstract: A Markov random field model is utilized to produce indices for residential real estate from repeat home sale data. A set of regions is represented by a graph in which neighboring regions are linked. This graph, repeated consecutively a number of times, with each region linked to the same region at adjacent times, defines a spatial--temporal graph that connects regions over space and time where each node represents a region at a particular time. An index is defined at each node to be the rate of appreciation of log sale price in a region during the preceding time interval. The indices are estimated from data consisting of repeat home sales. The Markov random field model specifies relations between neighboring indices, and between indices and individual repeat sales. A method is proposed for estimating various parameters in the model and for obtaining real estate indices. Following this prescription, biennial indices are calculated for the Dade County, Florida residential real estate market.
February 14, 1997
4:00 pm - 5:00 pm
116 Old Chem Building Any questions concerning the seminar may be addressed to Cheryl McGhee @ [919] 684-8029 or e-mail cheryl@stat.duke.edu. Please contact the author(s) directly for reprints etc.