Comparing asset and GPS sourced data
This whitepaper aims to understand and describe the effects of the COVID19 epidemic on workplace utilization in Quebec and how MySeat workplace data was able to capture trends that are similar to Google’s Mobility Report. This study determined that MySeat’s workplace utilization data gives results that are comparable to GPS data used by Google. MySeat’s data similarly captured a significant decrease in workplace activity although Google dataset also includes essential workplaces.
MySeats dataset consists of anonymous office utilization data gathered from different venues in Quebec that serve around 10,000 employees. These venues would be categorized as “workplaces” by Google for the province of Quebec.
We wanted to see what our own clients activities were in correlation with the GPS-based measurements.
At the very beginning of this pandemic, we reached out to our clients individually informing them they can use their data to observe the space utilization while their offices were being vacated, making sure the transition to working from home is effective and done in a timely manner. On an ongoing basis, the same data can be consulted to oversee when there is new activity on their sites and post-covid, when we can cohabitate once again, only prescribed to be at a distance.
Here in Québec
We learned about the existence of the Google mobility report while listening to the official speech of PM Francois Legault on April 8th. Quebec government leaders quoted the report to highlight that Quebec is ranking the highest in respecting social distancing measures compared to other provinces. Of course we wanted to see what our own clients were doing in correlation with GPS-based measurements. We were greatly anticipating that Google would do something towards measuring the effectiveness of confinement instructions and are excited to talk about the results of this comparison.
Excerpt “Google prepared a report to help their users and health officials understand responses to social media. distancing guidance related to COVID-19. Google mobility report shouldn’t be used for medical diagnostic, prognostic, or treatment purposes. It also isn’t intended to be used for guidance on personal travel plans. Location accuracy and the understanding of categorized places varies from region to region, so we don’t recommend using this data to compare changes between countries, or between regions with different characteristics (e.g. rural versus urban areas). We’ll leave a region out of the report if we don’t have statistically significant levels of data. To learn how we calculate these trends and preserve privacy, read About this data.”
Similarities in the results
Trend similarities can easily be visually identified as being closely correlated. The comparison shows the following results:
The overall patterns are similar (averaging a distortion due to known factors) and knee points of both curves have close timestamps (nearly synchronized) and they occur in the same chronological order for both data sets. We can recognize similarities within the areas highlighted by different geometrical figures.
The decrease of activity occurs earlier and faster in MySeat data.
Overall decrease over same time period
-44% Google // -97% MySeat
Note : Y-axis has been rescaled for both graphs for convenience of comparison
Decrease of activity occurred earlier in corporate workplaces than other types of spaces
MySeat reached out to our clients, giving them a heads-up on how high the office occupational levels were showing on our dashboards. This corresponds with a persistent activity in the (common) areas of their buildings, despite overall decrease.
Due to people finishing work on-site, lack of clarity of instruction from health official and local government, confusion between mandatory confinement and voluntary self isolation, both curves show a pattern of temporary come-back – that occurs twice on the Google data set, while it barely occurs for MySeat users.
Google mobility report certainly includes “essential workers”
Users for whom the workplace is a medical centre, a pharmacy, a factory, a grocery store, while MySeat users exclusively work in corporate offices and for a large majority can continue working from home. It is not surprising that all corporate workplaces record a drop that is twice as much compared to all other workplaces (from all categories).
Proportion of essential workers
According to our research, there are at least 600 000 essential workers in Quebec* that were invited to continue working. Is it safe to consider that a sample of these workers are responsible for the 46% remaining activities in workplaces, only Google can tell!
To pull a comparative data set from MySeat, we defined the date range for the baseline as defined by Google. from : 3-01-2020 to 06-02-2020.
Google’s definition of baseline: The baseline is the median value, for the corresponding day of the week, during the 5- week period Jan 3–Feb 6, 2020. The relative variation V for a given day J is calculated as follows:
Data plot over a period of 7 weeks from feb 16 to 29 mar 2020
MySeat Occupancy Score for a given day : S(J)
BL(J)) : Baseline for the corresponding week day to day J
V(J) : Relative variation for the day J
b(i,j) : is a % of time while the workstation WS was active for a given hour
Insert formulas here for BL, V(J), S(J).
Comparing Google trend data with MySeat
X-Axis are superposed
Y-axis have been recalled the right proportion to visualize and compare trends
Comparison : both graphs displayed are for a 7 week period of time : from 16-02-2020 au 29-03-2020
Privacy and data
While MySeat uses asset tracking – inherently anonymous – data for the analysis which is accessible in real-time to venue managers, privacy concerns have been raised by Google mobility report writers, who have chosen to discard certain areas due to small size of samples that can not “guarantee anonymity”. See full disclaimer on page 10 of this report:
For both Google and MySeat data, each data set is – foremost – representative of their respective users. While all MySeat current users work in offices, Google may also include users who work from factories or other places of employment. But we have to recognize that it is easier for an analyst to work with asset-oriented – ie anonymous – data to evaluate trends rather than using people’s private data via their GPS. This type of collection makes it slower for Google to release such reports, as explained on the last page of their report. Ethics suggest to discard GPS data when sample size is not deemed large enough to insure privacy. Asset tracking data is irrespective of personal data concerns, since it doesn’t use any, and the results are always reliable.
Since there are so many more connected mobile phones than connected physical assets, Google’s mobility report remains of a major general interest for its scale, while MySeat data can only be insightful for a fraction of territories that are being documented by Google. That said we certainly remain very insightful towards our clients, that are equipped with our technology.
This study showcases that social distancing can be evaluated using asset tracking data. Advantageously, it allows more flexibility to an analyst, who will not have to make the decisions on what needs to be excluded or filtered out to find the insight. Tracking assets is a simpler tool, requiring less time to strip down the otherwise enormous data set.
Expanding the Internet of Things will have to allow for humanity to gain more understanding of the population’s mobility without compromising their privacy. IoT designers must strictly pay attention to making anonymity at the heart of their designs in order for the population to be reassured as to its obscurity and embrace the utility.
Authored by Yahya El Iraki, Pouya Yousefi and Sandra Schmitke