"This graph here shows essentially that I've plotted acceleration thresholds on the bottom. Level one is a mild acceleration event, recorded by phone, and level five is a very hard acceleration event, monitored by a phone. And I've got the relative difference between different smartphone types.
If you left this unfiltered and unchecked, you would have a scenario where somebody would have an Android and an iPhone and do one journey and get two different scores. You have to normalize the data source. This will become increasingly important when we start getting data from other sources, like the connected car, like an existing black box that's been installed in that vehicle for whatever reason.
From activity monitoring services, from a smartphone, through an Android—you have to be able to treat all data, make it uniform and be able to use it regardless of its source, and that's the challenge we're spending time on.
Data can vary from up to 50% depending on the smartphone type." (Watch video for more...)