Identifying Occupancy Accuracy Issues

When an occupancy system has accuracy issues this will usually be evident by examining the occupancy data at various times throughout the day.

Looking at the data to see what happens first thing in the morning and last thing at night could reveal issues.

In the morning as people generally will be arriving to an occupancy area, the first 10 minutes’ worth of data should be mostly IN counts, and the very first count will always be an IN.
Conversely at the end of the day the vast majority of data should relate to OUT counts, and again the very last count should be an OUT.
If in your data, you see the opposite of this, this points to a simple direction issue which needs to be corrected either on device or in the direction assignment in the reporting software.

Similarly, if the data from multiple zones doesn’t look right, knowing the layout of a particular environment and following the first counts of the day as people move further into a building and between zones could show obvious directional issues. For example, if you enter a building and are considered in a zone, and you can then continue on and through to a second monitored zone, you should see a main door count, with an ‘IN’ to the first zone, then an ‘OUT’ of that first zone, then an ‘IN’ to the 2nd zone. If you see an ‘IN’ to the 1st zone, then another ‘IN’ to zone 1 followed by an ‘OUT’ from zone 2, this will indicate the unit(s) between zone 1 & 2 is configured the wrong way around.

The main time for data issues to be revealed are at the end of the day when an occupancy area is known to be empty. Note this only applies to buildings/shops/sites that have defined opening (or ‘office’) hours. It does not necessarily apply to a 24/7 site.

At the end of the day, after a full days’ worth of counting, in an ideal world where everyone is counted IN and OUT at 100% accuracy, the occupancy value would be reported as zero reflecting the empty/closed status of the shop/office/building. In reality, due to the inaccuracies that may have kept into the system during the day, reporting a value of zero will be rare, and instead, the calculated occupancy figure being reported will be a value higher than zero, or a negative value (dependent on the IN/OUT count bias). 

If the ‘end of day’ occupancy value being reported is a positive value, then this indicates an issue associated with higher IN counts than OUT counts. Inversely, if the value is negative, this indicates an issue associated with higher OUT counts than IN counts. If the occupancy value being reported at the end of day is very close to zero, this could be seen as a good sign, indicating a system with good overall accuracy, but be aware that it could simply indicate that the IN and OUT counting are both inaccurate by the same amount, so is not always the true story.

The only way to know that any device is performing with a high degree of accuracy is to perform proper validations comparing captured video recordings with actual counts over the same period.