Real Estate Analytics and Future Trend Prediction using Neural Networks and Map Personalization

Abstract

In 2015, the worth of global real estate was $217 trillion, which approximately makes 2.7 times the global GDP. This also accounts for roughly 60% of all conventional global resources, making real estate as one of the key factors behind any country’s economic growth and stability. Since location is termed important when it comes to real estate and consequent decision-making, digital maps have become an exceptional source for real estate purchases, planning and development. Personalisation can assist real estate users to make judgments by identifying any person’s desires and inclinations, which can then be recorded or captured as a user performs interactions with a digital map. This information can then be used by a personalised real estate map to suggest properties on the internet, assisting homeowners and providing useful real estate analytics. For users of a real estate platform, we have created a recommendation engine based on content, collaboration and location providing users with useful recommendations along with a house price prediction model, which employs the technique of multiple linear regression and neural networks. Our prediction model classified increasing, decreasing and stagnancy of pricing trends in different areas of Islamabad city. The results show 79% precision on recommendation and an accuracy of 80% on price prediction. This approach can be useful for Pakistan's real estate industry, with the primary goal of changing current management practises by employing GIS and data analytics as enduring solutions.