"The COVID-19 pandemic is without a doubt the largest event in recent memory to have disrupted the lives of billions of people on a global scale. Many across the globe have changed their daily routines: how they work, play, and otherwise spend their time." [1]
While the pandemic is now a constant in daily life, the ways in which society has adapted its mobility is ever-changing, with factors such as seasonality and changes in public policy directly affecting movement. In this interactive explainer article, we detail changes in mobility in Canada during the COVID-19 pandemic, with emphasis on the province of British Columbia, through a series of visualizations. "Furthermore, we will not be visualizing or generating any data on how COVID-19 progresses in a region as we are not experts in epidemiology, and presenting data in this space would be unethical." [1] Instead, our visualizations and data are centered on the following mobility categories:
The original source for our data is provided by Google's COVID-19 Community Mobility Reports; the data is presented as a percentage change compared against the baseline five-week period of January 3 - February 6, 2020, and begins on February 15, 2020. This data is collected and maintained by Google and is frequently updated with newer data.
While the data collected by Google is incredibly thorough, there are dates for which there is no reported mobility. This is because Google does not release information for a specific date or mobility category if there is not enough data to maintain privacy guarantees. Because the data for the baseline is static and collected in January, it does not take into account any expectations we would have for future mobility (e.g. seasonality). For example, we later observe that mobility in the Parks category is more than 200% times the baseline during the summer months. However, this makes sense with our knowledge that people generally do not go to parks much in the winter since it is cold, but will go more frequently when the weather is warmer. Furthermore, the data collected for the Residential category is measured as “hours spent” rather than total visitors. Therefore, we will see that there are small changes in Residential mobility since people will generally spend most of the day at home. Lastly, because mobility is reported as a percentage change from a baseline, we cannot draw any conclusions about the total mobility for any category.
In the following sections, we will present a number of visualizations alongside short explanations for the data and trends, but we leave further exploration to you, the reader.
Our first visualization provides a choropleth map representation of Canadian mobility. From this, we can see per-day differences and similarities across every province and territory. We also provide a play button if you wish to see an animated version of this visualization. With this animated version, we can already start to see some trends emerge in the data. If you are curious as to the exact values for a province or territory, simply hover over the region of interest and a box will tell you the exact percentage change. Regions with missing data are represented with a light grey.
While choropleth maps can show a lot of information, we should take note of their limitations. For example, what if we see a 50% increase above baseline in both British Columbia and Nova Scotia? Does this mean that we should view these results as equally surprising or should we be more surprised that a province with a lower population (Nova Scotia) has the same percentage increase in mobility as a province with a larger population (British Columbia)? To help answer questions like this, we present a Surprise Map visualization for the mobility data to show if there are any anomalies in interprovincial mobility.
Surprise Maps were originally introduced by Michael Correll and Jeffrey Heer as an alternative to standard choropleth maps [2] and allow us to see if there are any anomalies or points of interest that we should explicitly focus on. In this visualization, we present the mobility data as a signed surprise value rather than a percentage change in mobility. The signed surprise shows whether the given percent change in mobility is above or below the expected amount. Two prior models determine the expectation. The first model assumes that the mobility between provinces should be approximately the same. The second model assumes that the mobility in each province for each category should be approximately zero with the variance inversely proportional to the population for each category. Again, any dates where there is missing data are shown in a light grey.
One subset of the dataset to focus on is the summer months of July and August for the Parks mobility category. If we look at just the percentage change in mobility, we can see that mobility in almost all provinces is much higher than the baseline. However, when we look at the surprise map for the same corresponding dates, we see that only some regions have higher or lower values than expected when comparing between provinces.
Public transit is an invaluable part of everyday life for many people across Canada. In British Columbia alone, there were a total of 453 million public transit boardings in 2019. This is an increase from 363 million boardings in 2015. In fact, the designation of the busiest bus line in North America belongs to the 99-B Line in Vancouver [3]. The COVID-19 pandemic has led to restrictions on workplaces and social gatherings. Naturally, people are no longer travelling at the same levels they were before the pandemic. For example:
"Close to half of bus riders have not returned to transit as TransLink resumed full services this past summer after limiting them in the early days of COVID-19." [4]
and this is corroborated by the visualization we present on the right. This pattern is also not something that is local to British Columbia or one Canadian province or territory. Based on available data from February 14, 2020 to December 1, 2020, it is clear that most regions of Canada have not returned to the levels of travel observed before the pandemic.One may surmise from this visualization that, in general, people are not travelling as much as they used to. An interesting observation is that although regions were under heavier travel restrictions at different times, the general pattern of decreased transit mobility appears to be common among them throughout the recorded time period. It is also interesting to observe the almost immediate drop-off in transit mobility in mid-March. This can be explained by the imposition of harsher travel restrictions after the WHO declared COVID-19 outbreak a pandemic [5].
The workplace has undergone a number of transformative changes over recent years. From cubicles to open-office concepts, the modern workplace is, for better or for worse, the place where most people spend the majority of their waking hours. With the introduction of public health measures to curb the spread of COVID-19, many workplaces have “gone virtual.” This means that many people who would otherwise be commuting to the office for work are instead telecommuting where applicable. In the visualization to the right, one can clearly see the sharp dropoff around mid-March when COVID-19 restrictions around Canada were introduced.
Statistics Canada reported earlier this year that nearly one-third (32.6%) of businesses had 10% or more of their workforce were working remotely on May 29, 2020 [6]. The troughs in the previous visualization can be explained by workplace mobility understandably dropping toward the weekend. Major drops are explained by regional/nation-wide holidays such as Family Day or Canada Day. Looking beyond the direct impacts of COVID-19, it is interesting to consider the future implications that this pandemic has had, and continues to have on the future of work. Statistics Canada also reported that close to one-quarter (22.5%) of businesses expect that 10% or more of their workforce will continue to telework or work remotely [6].
An interesting observation to note is that this mobility category is one that demonstrates a trend in a positive direction. Since fewer people are commuting to work, attending social gatherings, and in general, leaving their homes less, we see that there is an increase in mobility for the residential category.
With most citizens being predominantly homebound since the pandemic began, many are looking for activities to both socialize and remain active in a safe manner. Weather-permitting, parks and outdoor spaces have become a hotspot for meeting friends at an acceptable distance, exercising, and taking a break from the new normal of home which now often includes work, school, and leisure.
In fact, CBC News recently reported that regional park use in Metro Vancouver has increased to 14.4 million visits by the end of October, a 37% change from the recent averages of 11 to 12 million per year [7]. A recent survey by Park People, an independent charity for building community through parks, found that 70% of the 1600 residents surveyed across the country have had an increased appreciation for parks and green spaces since the COVID-19 pandemic began. The respondents cited mental health, physical health, connection to nature, and social connection as top reasons for this appreciation; the latter was found to be particularly high for those who live alone. [8]
These patterns appear to extend to our visualization to the right. It shows large increases in park attendance from baseline for both British Columbia and Canada overall. However, in the national view, we note that the variation is less extreme. We partially attribute this variance to the decreased viability of visiting parks in the winter in many provinces and territories in Canada due to weather. As noted by Google, readers are reminded that park attendance will naturally increase in the warmer months, so some positive change from the baseline taken in January and February is expected.
Due to lockdown measures and general safety concerns, retail industries have seen a significant change in traffic during the pandemic. With most people spending more of their time at home, e-commerce, socially-distanced curbside pick-up, and local delivery have become immensely popular in comparison to standard storefronts. Many retail chains which rely on physical stores have gone bankrupt or converted to online-only storefronts in record numbers, [9] whereas online retail giants like Amazon have seen massive gains [10]. Additionally, likely due to both lockdown restrictions and safety unease, recreation industries such as sport have also seen major declines with all professional sports leagues taking significant breaks [11].
Experts at KPMG, a multinational accounting and professional services network, discuss how many retailers have seen a sharp decline in “physical footfall” at brick-and-mortar stores and malls during the pandemic and, during strict lockdowns, a complete closure of these locations. However, in contrast, they note that other essential retailers, such as grocery stores, have had an inverse trend and dealt with low stock due to stock-piling and sourcing issues. [12]
Interestingly, our calendar heat map may reflect this suggested trend of stock-piling goods as consumers appear to have visited grocery and pharmacy locations more than average in the few days before British Columbia implemented lockdown measures in mid to late March. We also see sharp, consistent decreases in retail and recreation mobility across Canada. As many experts suggest, the pandemic will very likely permanently shift the way we shop and otherwise spend our leisure time.
The previous sections were useful in observing changes in a given mobility category across a number of regions. It is also interesting to compare changes across a number of mobility categories while holding a geographic region fixed. For example, a person living in British Columbia may want to observe if there is an inverse relationship between mobility in transit stations and residential areas, which they may do so by selecting the respective categories. This visualization enables comparisons across mobility categories for provinces for which there was was enough data to make useful comparisons.
Pairwise and one-to-many comparisons between mobility categories are also enabled by this visualization. Some trends in the line graph become easier to observe and to interpret when they are coupled with the calendar heatmap for British Columbia. One example is the drops in mobility for workplaces in British Columbia for provincial and federal holidays, which are delineated in the calendar heat map.
Visualizations implemented in the Javascript D3 library using Observable Notebooks.
Code sources, including more customizable visualizations, available here:
Cover image: https://www.huntsville.ca/en/recreation-and-facilities/covid-19-facility-and-service-updates.aspx?_mid_=48819#.