Critical analysis and effectiveness of key parameters in residential property valuations
MetadataShow full item record
All municipalities are required to re-assess their real estate periodically which they do manually spending large sums. This research has developed statistical and AI models for such mass appraisals. Three statistical and AI models: multiple regression (MR), additive nonparametric regression (ANR), and artificial neural network (ANN), were developed using the housing database of the Town of Amherst with 33,342 residential houses. Prediction accuracies of the three models were checked and found to be acceptable, and the results of each of the three models were linked to the GIS map layer of the municipality to draw various maps showing the distinct and wide variations in the prices of homes based on location or neighborhood. The research confirmed that statistical or artificial neural network models are reliable and cost effective methods for mass appraisal of residential property values. The time variations of the housing prices and their interaction with the macroeconomic indicators: Oil Price ( OIL ), 30-year Mortgage Interest Rate ( IR ), Consumer Price Index ( CPI ), Dow Jones Industrial Average ( DJIA ), and Unemployment Rate ( UR ), were analyzed using Vector Autoregression (VAR) on the monthly housing sales data for the Town of Amherst, State of New York, for the period: 1999 – 2008. The various analyses concluded that the 30-year mortgage interest rate ( IR ) has the highest effect on the housing prices progressing from 4.97 percent in the first month to 8.51 percent in the twelfth month. The unemployment rate ( UR ) was next in order followed by Dow Jones Industrial Average ( DJIA ), and Consumer Price Index (CPI). This research, also finalized a methodology for dividing the housing stock of a municipality into uniform zones or districts for value assessments and other planning purposes. The cluster analysis was utilized to regroup the existing 67 neighborhoods of the Town of Amherst into 25 districts to simplify the priori classifications. The methodologies formulated and tested in this dissertation will be useful for municipalities and consultants while performing mass appraisals. Town planners will also find the various methodologies and the resulting patterns useful for determining the development needs of one district in comparison to all other districts.