Microsoft has partnered with the Federal University of Minas Gerais, one of Brazil’s largest universities, to undertake research that could help predict traffic jams up to an hour in advance.

The Traffic Prediction Project is setting out to crunch all traffic data, including historical numbers where available, from transport departments, road cameras, Microsoft’s Bing traffic maps, and even drivers’ social networks, to see if established patterns can help foresee traffic jams from 15-60 minutes before they happen.

Big data is increasingly being used to analyze global problems to find solutions; this extends into the medical realm too, where it’s being used to discover new drugs and even combat health care fraud. So beating traffic jams is just one of many real-world issues that can be tackled by combining lots of data from multiple sources.

While there are a growing number of tools and online services that can show drivers congestion hotspots in real time, including Google Maps, it’s often too late given that a driver may well be approaching the bottleneck already. Being able to accurately predict jams before they happen has yet to bear much fruit, though many companies have been working on solutions.

In 2014, it’s estimated that 54 percent of the planet’s people lived in cities, up from 34 percent in 1960. This is expected to grow at almost 1.84 percent a year until 2020, then 1.63 percent until 2025. The growing urbanization of the world’s population means that whoever cracks the traffic jam-prediction nut will be onto something lucrative, with drivers able to take alternative routes, use public transport, or simply stay at home.

Microsoft will be putting its Azure cloud-computing platform to the test for the project, which will be necessary for the immense processing power needed to crunch multiple terabytes of data.

The computing giant says that it has tested its model in London, Chicago, Los Angeles, and New York, and claims to have achieved a prediction accuracy of 80 percent. That figure is pretty good on its own, but when you consider it was based only on traffic-flow data, it could rise to 90 percent when other data sources are thrown into the mix.