Predicting House Prices Using Multiple Listings Data

摘要: It is often necessary to accurately predict the price of a house between sales. One method predicting values use data on characteristics area's housing stock estimate hedonic regression, using ordinary least squares (OLS) as statistical technique. The coefficients this regression are then used produce predicted prices. However, procedure ignores potentially large source information regarding prices—the correlations existing prices neighboring houses. purpose article show how these can be incorporated when estimating and practical difficulties inherent in technique called kriging discussed. concludes with an example multiple listings from Baltimore.