SciTech Roundup, May 12

Using Twitter to predict traffic

With the power of big data and language analysis, Associate Professor of Civil and Environmental Engineering Sean Qian and his Ph.D. student Weiran Yao were able to use information scraped from Twitter to accurately predict morning traffic patterns. By applying language analysis models to anonymized, geotagged tweets, Qian and Yao were able to gain insights into the sleep-wake patterns of individual commuters as well as planned and accidental traffic incidents which affect congestion times. Even simple personal tweets can provide useful information, showing just how powerful abundant data can be when properly harnessed and interpreted. This new method helps supplement the lack of traffic information during nighttime, and can help transportation agencies better optimize their routes.

Read more about it here.

Vehicle electrification and ridesharing services

Companies like Uber and Lyft have changed the way we commute, with ridesharing services now making it easier than ever to get a ride using only your phone. However, mobilizing massive fleets of automobiles comes with an environmental cost in the form of air pollution, something the free market does not correct for on its own. In a study by Professor of Engineering and Public Policy and Mechanical Engineering Jeremy Michalek and his Ph.D. student Matthew Bruchon, it was discovered that if companies like Uber and Lyft were properly taxed for their contributions to air pollution, they would be incentivized to reduce that cost, and increasing electrification of their vehicles is one way of doing just that. While companies like Uber and Lyft are making attempts at increasing electrification, the study finds that the rate of conversion to electric vehicles would be greatly accelerated with properly imposed penalties.

Read more about it here.