Frotcom Analytics determines that a trip is a probable outlier by comparing it to all the previous trips that started and ended in the same spots.
Classifying a trip as an outlier requires Frotcom Analytics knowing some statistics about the previous trips. Using these statistics, Frotcom Analytics estimates limits, maximum and minimum, inside of which the trip’s values must lie in order for it to be considered “normal”. If one of the trip’s measures (e.g. distance) lies outside the estimated limits, then the trip is marked as a potential outlier (shown in the list with the Unknown type).
Note that Frotcom Analytics does not tell you that a trip is an outlier. It only points out this potential outlier, for you to analyse and in the end confirm or not as a true outlier.
To do that, you can inspect both the trips and the calculated detection statistics, on the Sample Trips pane shown below.
As previously noted, some trips may be classified as outliers on more than one measure. The picture above depicts a case where the trip was classified as an outlier both on distance and driving time, as you can see by the two links showing on top of the graph. In other words, the trip reported a larger mileage than expected and also took longer to execute than expected.
You can select under which dimension you want to inspect the sample data, using the selector at the top of the pane. As a suggestion, start by focusing on the distance outliers, because those are easier to understand by looking at the trip’s route on the map; usually there is a clear detour from the expected path.
In the Chart tab depicted in the image above, you can see a histogram with the sampled distance values. You can see two vertical lines: the Average distance line and the Maximum Limit line, above which all values are considered to be outliers.
Farther to the right, you can see the trip outliers’ distance observation in red, with a callout displaying the trip distance and the difference to the mean, in parentheses. This is an instance of an outlier whose value is larger than the maximum allowed.
The density line is a smooth representation of the histogram information, and may help interpreting how the data behaves. You can also manually change the number of histogram bars by checking the # bars and using the slider.
Detailed information about the sample trips can be found in the Sample Data tab:
Here you will find details of all the trips that were used to calculate the outlier statistics (the km column values, in the previous case).
The list footer is split into two sections:
- To the right you find summary information about the sample trips (average values, in this case);
- To the left there is
- A button that controls what is calculated by the summary (minimum, maximum or average)
- Two counts: the total number of trips, in black, and number of trips with CANBUS fuel information, in blue and parenthesised.
The checkboxes on the list row headings select the trips which will be displayed on the map to the right. Using this mechanism you can add as many trips as you like to the map to help you analyse and compare the outlier.
According to the selected measure, one trip will be selected as the “best trip”. This trip is automatically selected and displayed in bold on the list. The best trip is calculated using the same procedure used to find the outlier detection interval and in this case the first trip that has a value higher than the interval’s minimum is declared as “best”. Note that this calculation depends on the selected measure.