Real estate transaction prices are recorded and published for the previous 5-year period on the open data government website in France. This publicly accessible database provides data pertaining to the location of a property, both geographically and administratively, and to the property's characteristics such as price, area or type.
As a geographic data visualizing exercise and to genuinely help the apartment search of yours truly, this project first creates a Parision subset from a national database then proceeds to first visualize average prices on a street-by-street basis, then develop a multilayer neural network to predict "reasonable" prices based on property characteristics and location. As an extension to this project in the scope of an interview process, this project also proceeds to scrape different apartment renting websites to obtain average rent prices - limited to the Paris region - in order to help guide decision-making in real estate investments.
To see code and results, follow the link to my Github page.