Today cars are equipped with electronic devices with GPS locator (black box) which allow to record all the information coming from the various sensors of the vehicle and the behavior of the driver.
In this context, an important international company of telematics and IOT solutions for the Automotive sector requested a study to identify an improvement of the solution for the Crash Validation service for Insurance purposes.
A study was conducted on a Big Data/IOT dataset containing reports of probable vehicle accidents from devices in vehicles. The activity was carried out with the collaboration of the Incident Management and Risk Management Business Unit.
The solution made it possible to classify incidents into True and False, through the competition of Classification algorithms. The best algorithm with a high accuracy index was identified and at the same time had to respond to the primary objective of improving the current service, leading to a better reduction of False Positives and False Negatives.
The main benefit was the reduction of the impact of triggering the insurance process of the vehicle on incorrect reports of accidents, since the reporting of the Device generates a process of validation of the accident (Crash Validation) and consequently the activation of insurance services related to the vehicle.
Implementation of the Machine Learning Pilot service, from the creation of Machine Learning Models to Advanced Analytics.