Using Web Mining in the Analysis of Housing Prices: A Case study of Tehran

Abstract

There have been many previous works to determine the determinants of housing prices. All of these works relied on a relatively small set of data, mostly collected with the help of real estate agencies. In this work, we used web mining methods to generate a big, organized dataset from a popular national brokerage website. The dataset contains structural characteristics of more than 139,000 apartments, alongside their location and price. We provided our full dataset for the article, so that other researchers can reproduce our results or conduct further analyses. Using this dataset, we analyzed housing prices of Tehran in order to identify its major determinants. To this aim, we examine the dynamics of housing prices at the district levels of Tehran using Hedonic Price model. Our results highlight a number of points, including: Base area of an apartment is positively correlated with price per square meter (r=0.89), showing a two-folded impact on the overall price. Air quality is in a positive, and floor level is in negative correlation with housing prices.