Aussie researchers turn to AI for real estate valuations

Source: Xinhua| 2019-10-03 17:00:32|Editor: Li Xia
Video PlayerClose

SYDNEY, Oct. 3 (Xinhua) -- Assessing the price of land and property values can often be a time-consuming, labor-intensive process that results in costly, slow and subjective valuations.

But a breakthrough collaboration between the University of New South Wales (UNSW) and a range of government and industry partners in Australia, could revolutionize the way real estate assessments are performed across the world.

Using artificially intelligent (AI) systems, Value Australia compiles research and data analytics that can deliver accurate and efficient valuations for a broad range of land and property types without using a person.

"There will be a public-facing version of the software using a map interface to show property values around where people live, which will take into account land values and the amenity of an area," UNSW Professor involved in the project, Chris Pettit, said on Thursday.

"It will assist in making data-driven decisions, such as the ability to explore re-zoning options, determining the economic benefits of a development proposal, or even the location of important infrastructure, like new metro or light rail stations."

Building on a successful two-year pilot program across metropolitan Sydney and Brisbane, the project developed a number of Automated Valuation Models (AVMs).

Compared to market sales, the best of these AI models showed an accuracy level of 94 percent.

As well as making it easier for real estate agencies and regular people to find out the value of their homes, Pettit said the technology also has the potential to help investors, governments and businesses plan for multiple future scenarios, such as how forthcoming infrastructure projects may affect property prices in the area.