Buying a house or flat is the single biggest lifetime purchase for most people.
For many people it is a frustrating and disappointing experience. Estimates given by agents are often way out and many people invest much time and emotion on a property only to find that the final selling price is way out of their price range.
The realas.com app uses data from a range of sources about properties that are for sale to predict the selling price. Most of the predictions are within 5% of the final selling price.
In 2012-3 the ECML research group carried out an investigation to determine how accurately selling prices could be predicted from the available data. The results were surprisingly good and led to the launch of the realas.com web site in July 2014. The Realas start-up was acquired by the ANZ bank in 2017 and is currently, 2019, the bank's primary source of property price estimates.
The ECML research group is providing the data mining and machine learning expertise to drive the realas.com web site, monitor the prediction accuracy and update the prediction models as necessary.