Automobile companies recognized the need to understand how customers use their vehicles. To this end, telediagnostic modules were installed on fleet vehicles, with the consent of the owner, to collect and store usage data. This data was uploaded to a server when the module was able to communicate via Wi-Fi.
The volume of data is enormous. The size of a single vehicle file can be over six gigabytes, contain millions of records, and contain hundreds of millions of measurements. Each vehicle needs to be analyzed and the results from different vehicles combined to determine typical and extreme vehicle usage. With hundreds or thousands of vehicles to be analyzed, the analytic task is daunting.
To analyze continuous data, like speed, frequency histograms have been used. It will be shown here that the cumulative percentile histogram provides better information and can display single and multiple vehicle usage patterns. However, the cumulative percent histogram becomes too busy when many vehicles are included on the same plot. The data can be processed to determine the typical and extreme vehicle usage, which are then displayed on the cumulative percentile histogram.