The Data team at Zendrive announced an update from Algorithm core 1.2 to Algorithm core 1.3 today. What does this mean to our customers? Will driver scores suddenly change?
In this post I’d like to explain what is so exciting about the ongoing improvements we make to the platform and what it means to our customers.
Let’s start with an overview of one important feature of the Zendrive analytics platform. This story begins with the concept of algorithm defined event detectors. Whenever raw sensor data is transformed into an event, a detection algorithm is at work. Using the ever improving, ever growing sensors on your phone, Zendrive can understand deeply the quality and safety of a ride by evaluating detailed sensor data to build an event profile, and then comparing that event profile to thousands of other drivers on similar trips.
It takes humans many years to develop their brains to the point that they can respond to complex events appropriately. We even continue to improve our detectors for years, as long as we are presented with new cases to learn from. The Zendrive event detection algorithms improve in a similar manner, although in a much more rapid pace. They have been trained on tens of thousands of events to learn how to correctly identify important safety events from smartphone sensor signals. Yet we haven’t stopped there. As we continue to improve our libraries of known cases to test algorithms against, we can generate more and more effective event detection algorithms and deploy them. That is where the beauty of the Zendrive server-side detection technology really shines.
A simpler, less accurate, system would be hard-baked into an app or an SDK, and if the authors improved it, your users would need to get a new copy, and only their most recent events (those detected since the upgrade) would be detected with the new algorithm. That means that all of your historical data would be outdated. Not so for Zendrive Analytics. Our tireless data scientists and engineers created a system that can update your driver’s entire history of events to the current level of accuracy without missing a beat, or requiring you to press a button.
Here is how it works. The Zendrive platform pulls driver’s sensor signals and puts them on our server. As soon as a trip’s data has been uploaded, our algorithms crawl through the signals identifying events. A few seconds of activity later and your trip scores and events are sent back to you via the Zendrive API or Zenfleets dashboard. When we update our algorithms, we also update the scores back in time to make sure that all of your data is as up-to-date as possible.
Our team is constantly collecting new driving cases, and testing new algorithms. When an update is significant enough to earn a new name, we will release it to you. We hope you enjoy always-best-in-class event detectors from Zendrive.