Abstract
Single-family houses are typically the most important component in their owners' portfolios. Buying a home is a major financial transaction for most households. Yet, unlike assets traded in financial markets, getting a quote for the current market value of a house is not an easy task because houses axe very heterogenous assets. They differ, among other things, in size, location, age and maintenance. In this paper, we describe a web-based, almost realtime prediction of prices for single family homes in Berlin, Germany. Based on an extended hedonic regression model and estimated from a rich data set covering all house transactions in Germany's capital, this online service delivers predictions for homes with user-specified characteristics. We describe the statistical model and how its predictions are implemented on the computer to allow seamless interaction between its users and the data base containing the model estimates.
Original language | English |
---|---|
Pages (from-to) | 449-462 |
Number of pages | 13 |
Journal | Computational Statistics |
Volume | 18 |
Issue number | 3 |
Publication status | Published - 2003 |
Keywords
- hedonic regression
- Kalman filtering
- Java Server Pages
- REGRESSION
- ALGORITHM
- MODELS