Assessing the Rental Value of Residential Properties: An Abductive Learning Networks Approach

Assessing the rental value of residential properties is a complex and challenging process to both practitioners and academicians because it involves analyzing the rental property, neighborhood characteristics and market conditions. The rental housing market is characterized as imperfect and inefficient, because the product is long-lasting, fixed on a given site, heterogeneous, and controlled by extensive governmental regulation. Since each rental housing market is confined to a given area, characteristics of a market in one area are not necessarily an accurate representation of other markets. The purpose of this study is to build a model that can provide an accurate way of assessing the market value of residential rental property and analyzing the factors that determine market rents by using an artificial intelligence technique. Our main concern is to examine whether the Abductive Learning Networks (ALN) technique can be used to overcome many difficulties associated with the multiple regression technique which has been used primarily to analyze the price behavior of rental houses in the current literature (Guntermann, 1987; Murphy, 1989; Meacham, 1988; Jud and Winkler, 1991, Weirick and Ingram, 1990). The multiple regression technique is parametric and requires the user to specify the functional form of the solution. If one does not know or cannot guess the correct underlying form of the functional relationship, the regression approach will result in inaccurate models. Although the regression approach allows the use of a very general polynomial equation when the functional relationship is likely to be nonlinear, it becomes virtually impossible to estimate all coefficients, primarily because the number of coefficients to be estimated grows factorially as the number of variables and degrees of the function increase. Although the regression technique also allows the use of various transformations such as the Box-Cox (1964) transformation on the given data to adjust THE JOURNAL OF REAL ESTATE RESEARCH 1