Over the last few weeks, we’ve highlighted the amazing research and insights that have come out of SafeGraph’s COVID-19 Data Consortium: over 3,000 researchers, academic institutions, government bodies, and non-profit organizations are using SafeGraph data to fight the COVID-19 pandemic in real-time.
The research conducted by students at Stanford’s Future Bay Initiative is a perfect example of this. The team, mentored by Derek Ouyang, lecturer at Stanford University, tapped into the power of SafeGraph data to provide “actionable intelligence”—in other words, using data to identify the drivers we can change in order to arrive at the outcomes we want—around how local governments can approach their COVID-19 response efforts.
Much of this work has focused squarely on helping the City of San Jose (California) develop a more accurate and informed approach to dealing with COVID-19 head on.
The Stanford team first looked to SafeGraph data to identify noticeable patterns happening within the city’s limits. The end goal was to help the city identify where more compliance enforcement may be necessary to stop COVID-19 in its tracks.
Ouyang recently presented the outcomes of this work in an engaging webinar hosted by SafeGraph. It’s definitely worth watching on replay; however, if you’re pressed for time, here’s a quick overview of what the team has been able to accomplish.
Using a number of R techniques, the Stanford team brought SafeGraph data to life visually, helping the City of San Jose understand, at a quick glance, how well its population was complying with social distancing measures. This was made even more effective by visually comparing the data before and after stay-at-home orders went into effect on March 16, 2020.
While it’s no surprise that the city experienced a sharp drop in people leaving their homes immediately after the arrival of stay-at-home orders, the data also hints at the beginnings of a very gradual return to normal levels in May—likely continue in that upward trajectory as stay-at-home orders become increasingly relaxed over the coming weeks.
Even more interesting, when looking at the Weekday vs. Weekend comparison (see chart below), an uncommon trend started to play out: by early May, more people were leaving their homes over the weekends than during the week. This likely has a lot to do with the fact that many people, who are now forced to work from home, are using the weekends as an escape to get out for a breath of fresh air during non-working hours. Under normal circumstances, we would expect the percentage of people leaving their homes during the week to be higher purely based on work commutes alone.
The City of San Jose, specifically the Parks and Recreation Department, also wanted to know where people were going most whenever they left their homes. And given that many businesses were now closed, they realized that parks and open spaces could soon see a spike in traffic.
They needed this information about how visits to parks have changed in response to stay-at-home orders to help best allocate limited resources to enforce compliance with social distancing practices. This is especially important for suburban playgrounds, as those could quickly lead to increased virus spread if social distancing practices aren’t respected.
The Stanford team used SafeGraph Patterns weekly data, in relation to the geometries of San Jose’s park spaces themselves, to create a visual representation showing how visits (per acre) to those parks have changed compared to average foot traffic. These insights are available at the macro (all parks) and micro (individual park) level. See charts below.
Interestingly enough, this led the city to ask another important question: how many non-compliant, non-essential businesses were still in operation after the stay-at-home policies went into effect?
Unfortunately, using the same methodology proved to create a few false positives. After doing simple Google searches to understand where certain POI sit in relation to each other, it was found that many of the non-essential businesses still showing high foot traffic often shared walls or were in close proximity to essential businesses, thus throwing off the accuracy of the insights.
Even though the results of this analysis were inconclusive, at least in terms of being able to identify without a shadow of a doubt which non-essential businesses were not being compliant with stay-at-home orders, it nonetheless led to a couple positive outcomes:
Although this was not a direct ask from the City of San Jose, one of the big questions that the Stanford team has tried to answer is around understanding the relationship between various social distancing measures and virus spread.
More specifically, they were interested in understanding how certain community characteristics (demographics) connect to movement patterns as well as how these movement patterns may relate to virus spread.
One interesting example shared was a simple regression analysis looking at the percentage of people leaving their homes, before and after stay-at-home orders went into effect, in relation to the percentage of households with an annual income over $125,000.
Comparing the before and after begins to tell an interesting story: higher income households were more prone to leaving their homes before stay-at-home orders but also were found to leave their homes significantly less after those policies went into effect. While there are a number of ways to interpret these insights, what’s most interesting here is how data is used to identify what potential community characteristics may lead to increased movement patterns, which, in turn, may connect to virus spread.
With an eye turned towards the future, the Stanford team also used SafeGraph Patterns data to test the potential efficacy of contact tracing as a viable solution for reducing infection rates once implemented at scale in big cities like San Francisco.
More specifically, they wanted to measure how movement patterns can enhance our understanding of basic variables around disease spread and contact tracing—with the end goal of developing more accurate simulations around the impact of different intervention methods in mitigating the spread of COVID-19.
The data was used to calibrate unknown parameters in virus spread rate models and then validate those modeled processes by comparing the outputs to SafeGraph Patterns data.
The big takeaway here is simple: the SafeGraph COVID-19 Data Consortium is helping national, state, and local governments approach mitigating the spread of COVID-19 in a number of unique, innovative, and potentially life-changing ways. The work from Ouyang and his team is just one of many examples of how SafeGraph data is being used to solve real world problems.
Be sure to take a moment to watch Ouyang’s webinar on replay. You’ll be glad you did.
Contact our team today to learn how these datasets can help your business navigate this unfolding health and economic crisis.
