Frequently Asked Questions


Q. What's considered a "week" on Benchmark > Periods?

A. Weeks are calculated from Monday to Sunday and are identified by their first day (Monday).


Q. Which vehicles are taken into account when making the calculations on Benchmarks?

A. Benchmarks are calculated using trip information. All vehicles that produce trip information to the system will be used in the given period.


Q. How's the average workload calculated?

A. The average workload is calculated using the working days (see settings dialog). For the working days, the average considers the whole 24 hours of the day, and only the amount of time accounted for each trip is considered.


Q. Which information is used to determine the Vehicle profile (heavy, light).

A. We have implemented a machine learning algorithm that leans trip behaviors from pre-specified accounts (known accounts with heavy or light vehicles only) and classifies vehicles with at least two months worth of trip data in the system. This algorithm only classifies vehicles that have no set Vehicle class on the Administration section of Frotcom. For those, we have a simple mapping correspondence (e.g., Truck = Heavy, Car = Light).


Q. Which information is used to determine the Haulage type (long, short).

A. We use a heuristic to classify haulage: taking the vehicle's last 30-day worth of trips, we calculate the geographical bounding box for all points. Long haul vehicles have a bounding box diagonal over 400 km. All others are considered short haul.


Q. How many trips should there be between two places to see them as Frequent trips? What about Trip outliers?

A. We only consider at least ten trips between spots (spots are a definition in Analytics for locations where vehicles frequently stop). Same goes for outliers.


Q. How many stops should there be in a place/address, and for which period, to see it in the Place editor.

A. Places are created in Frotcom, so they do not depend on any detection. On the other hand, Spots are detected using a density-based clustering algorithm (based on DBSCAN), and we consider only clusters with at least ten vehicles "density-reachable" using a link distance of 25 meters. Please note that some large clusters can be further segmented using half that distance. Large clusters are classified using a heuristic rule.

Was this article helpful?
0 out of 0 found this helpful
Have more questions? Submit a request


Powered by Zendesk