Last week, Lyft accused Uber associates of requesting thousands of rides from its drivers and canceling them at the last second–a claim that Uber said was “patently false.” In response, Uber (which is much larger than Lyft and operates in three times as many markets) started hurling allegations that Lyft was doing the same thing and was even seeking a potential acquisition by Uber.
Messy? You bet. But now a data startup called Linkurious has put together a theoretical–but nevertheless nifty–visualization that might help some people make sense of the Lyft vs. Uber feud. The goal is to help figure out whether marching orders to cancel rides from either service may have been issued. In order to do that though, first you have to figure out what a typical user account looks like.
“I have pulled up this data in Linkurious, the graph visualization tool we develop to understand what the people at Uber [or] Lyft may have experienced,” writes Linkurious cofounder Jean Villedieu in a blog post about visualizing fraud patterns, and what data scientists may have looked for when sniffing out abnormal activity. “When investigating a large dataset identifying patterns is key.”
Agents of both services were reported to have summoned and later canceled rides using multiple registered accounts. Working off news reports, Villedieu plugged in the data for 30 hypothetical rides, with (fake) names, IP addresses, and rides summoned/canceled that Lyft and Uber would have used to identify sabotage. Here, for example, is what a normal user profile for either Lyft or Uber might look like:
One name attached to the account. One IP address. One phone number. Six rides taken, and one canceled ride. On the other hand, here is what one IP address from a mobile device with three different names attached to it looks like:
Three different names. One number. One IP address. Fifteen canceled rides. It’s a very different-looking, multi-tentacled representation, especially compared to what a “normal” ride might look like. Quickly identifying these might help data hounds at both Uber and Lyft unravel potential cases of fraud.
Head over to Linkurious to check out the rest of the data–and to see what a slightly stealthier agent might have done.