Estimating Housing Price: Kriging the Mean

The price of houses and its evolution in recent years is one of the issues that citizens and, of course, political and economic authorities are worried about. In most of the developed countries the mean price (per square meter) of houses in an area is estimated by a simple average of single prices from a sample of houses that does not take into account the spatial correlation among the prices. As an alternative to this classic procedure, in this paper we propose a linear estimator of the mean price of houses, using the kriging estimator, which has been specially designed for the case of spatially correlated data in a given domain. This estimator is the best unbiased linear one and provides a more realistic estimate of the mean price of houses in the urban area we are interested in. Obviously, the modelling of the variogram function is a central point in the global estimation process.