Introduction
This article provides answers to some of the most commonly asked questions that we receive and common issues customers face in interpreting insights.
You may also find these articles useful:
I’m trying to reconcile a complaint to a graph...
Customers often need to validate a complaint against performance monitoring in Mooven. Most of the time, this works smoothly and there is clear evidence in line with expectations; however, sometimes you can get a curely question from a trustworthy or influential source that you can't answer easily.
What could the issue be?
Variability in individual experiences:
We often talk about the typical experience of road users, representing the 50th percentile or median journey time. In practice, there can be a significant difference in personal experience depending on the timing or driver behaviour.
In the example below, we can see journey times for individual vehicles while roadworks were active, captured using ANPR cameras.
Looking at journey times between 10 am and 4 pm we can see clear vertical bars, indicating the there was a wide spread in journey times from the cars that were first in a stop/slow queue to the cars that joined the back of the queue and then could start moving quickly again.
In this example, the journey times varied from 25 minutes to 55 minutes on the same road. While the 55-minute journey time was a real result, the median travel time was 32 minutes, showing that a complaint may be founded but not representative of the experience of most motorists.
Timing of the complaint versus data capture:
Where customers are looking to be very precise with timing, another common situation has to do with how data is aggregated into 15-minute time buckets.
In the below scenario, there is a short spike in delays above 5 minutes which has been recorded in the 8:00 am time bucket.
If the complaint was received at 7:57 am about a 8-minute delay, it is possible that the next data point was only captured at 8:07 am, ten minutes after the reported event. As the 8:07 am data point relates to the 8:00 am time bucket the 5 minute delay is shown against 8:00 am. It is highly possible that delays had reduced from 8 minutes to 5 minutes over this time period.
I’m trying to reconcile an alert to the graph...
This is the same question as the one above. Our alerts are triggered as soon as we receive a response from our data provider, while the graphs show information aggregated into time buckets.
I have two routes that are showing different results...
There are several reasons why two routes covering a similar section of road might yield slightly different results.
Different sites: If the routes are associated with different sites, the sites might be polling data at different times.
Variations in route length or design: If one route goes through an intersection, while the other one stops 50 meters before the intersection the travel times will be different. In the first example, this route will include delays associated with the intersection.
Different data sources: The routes may be using different datasets. For example, one person might be using TomTom, while the other route might be polling Google.
Time shift effects: If you are trying to sum segments and compare them to an overarching segment, the results may not always match. This is most likely due to time shift effects. For example:
You have four routes of 5 km each and one overaching route at 20 km.
Each data request will provide the current journey time if you were to complete the segment now; however, for the 20 km route, it would take time for the driver to reach the 15 km mark. As the location and timing of congestion can move, you will get different results as time progresses, depending on your location.
I used my mapping provider to provide directions to work, and they got it wrong. How can I trust floating vehicle data?
Mapping providers have invested vast sums of money into providing the most accurate journey time possible - and they do a remarkably good job most of the time, considering how many things can change during a journey.
One common situation that occurs, which will result in incorrect estimates, is unexpected incidents. When considering a long journey in rush hour traffic, likely in the image below, the original calculated journey time may change materially if an accident occurs 20 minutes into your journey. In this case, there is nothing that companies like Google or TomTom can do to predict an incident, and we wouldn't want them to muddy actual observed data with predicted data.
Another common situation that can cause differences is either extremely slow (<5 km/h) traffic or multiple lane behaviour as outlined in the first example. When speeds drop to 2 or 3 km/h, small variations in speed can often result in significant differences in total journey time calculations. We have seen a difference of 10-15 minutes or more, due to a variance in speed of less than 2 km/h.