Kannattaakokauppa.fi - Asuntojen hintatrendien tilastollinen mallinnus ja visualisointi
Reaktor - Published 1.2.2018 - updated 3.3.2020
Kannattaakokauppa.fi visualises apartment price trends by postal code regions in Finland since 2005. With the help of Bayesian probabilistic modelling, we can get much more reliable estimates of the real price and trend values than by looking at individual data points. The results are provided in a nice interactive visualisation that allows the exploration of prices and trends across the country. The service also provides future predictions.
Regional apartment prices are currently reported annually as simple lists of the top and bottom regions. Such lists are based on very noisy values, especially in regions with very few transactions per year, and can thus lead to incorrect conclusions. This is an example where statistical modelling can help to provide more reliable conclusions.
In the future, we would like to include more postal code level data to better explain the price trends. For example, demographics data from Paavo could be integrated into the model, so we could both provide better estimates and also identify which demographic factors are related to apartment prices.
Details of the data and statistical modelling can be found from the links below.