Coaching Reports

Concept

This feature is dedicated to companies which have review service paid and its purpose is to assign a label or more labels to a driver for his driving behavior in last week.

Coaching Drivers Categories

There are 6 distinct categories in which a driver can be assigned into. A driver can be simultaneous in multiple categories (except if it a good driver, in that case, that will be the single label).

In the table below are the reviews that are taking into consideration into algorithm for each of the labels:

• The report takes into consideration last 12 weeks, a period in which we can see the drivers’ evolution in each category;
• Each category will have a top 10 drivers for the current week and the possibility to see the entire list by pressing ‘Full list’ button;
• These are weekly reports, so the period is limited to weekly format. On each Sunday, a cron analyses events and clusters drivers based on them.

Important Notice

‘Fuel Wasters’ category is available only for clients with cellular data. (to calculate wasted money, we need ‘live’ data, meaning speed at each 3 seconds – now, this data is sent each 2 minutes, when camera sends the data package).

Let’s take, for example, ‘High-Risk Drivers’. We can see:

• The evolution of the reckless drivers in the last 12 weeks - represented by the graph;
• How much these reckless drivers represent from the total number of drivers – represented by the pie;
• How many weeks from the last 12 weeks that particular driver was assigned in that category – represented by the blue bar and the number. (this means that Emmanuelle Torres was a reckless driver for 11 weeks in the last 12 weeks).

Let’s take, for example, ‘Fuel Wasters’. If you click ‘Full list’ button you can see:

• A scrollable list with all the fuel wasters in the current week;
• The evolution of the drivers from this category over the period of the last 12 months;
• The location of each driver;
• A search bar to find a particular driver;
• A sum in the right side of the driver name that represents an estimation of how much extra cost that particular driver is making per mile per week (by idling and speeding).

Algorithm

1. Collecting data from database: drivers, company and review id of events only for drivers who drove more than 5 miles in last week.

2. Collecting drivers’ total distances made last week.

3. For each driver is calculated the frequency of occurrence for all event types from database which is divided to square root of driver’s total distance named event occurrence per distance.

4. Forwards is calculated a threshold for each event type based on event occurrence per distance as events’ mean + 1 events’ standard deviation. So, each event type has its own threshold which is dynamic (it changes weekly).

5. Afterwards, is calculated the difference between each event frequency and event threshold for all drivers.

6. If the event outcome is greater than zero then is calculated the sum of all events which compose every drivers’ category (Speeders, Reckless Drivers, Distracted Drivers, Traffic Offenders). If the event outcome for each event type is less than zero, then he will be classified as a Good Driver.

7. Drivers will be assigned to the groups for which they have the sum of outcomes greater than zero. Thus, a driver could be classified in one or more groups.

8. For each driver is calculated the fuel consumption while making idling and while making speed events.

9. For every driver which have the fuel cost above average + standard deviation of the entire population, the label assigned will be ‘Fuel Waster’.