data visualization
Compare how various industries were impacted by COVID-19 and track their progress.
The coronavirus (COVID-19) global pandemic brought massive and ongoing changes to how people interact with their surroundings and participate in the economy.
SafeGraph, committed to open access to information, created a dashboard to share insights into economic health, and how America adjusted to a new normal. This dashboard covers the time period of January 2020 to May 2021 to show how the effects of an unprecedented year.
This map and dashboard are built from SafeGraph Patterns: a dataset of foot traffic counts and visitor insights for millions of points of interest (POIs). With data on thousands of retail chains, including smaller, mom-and-pop businesses, this data provides unparalleled insight into US industry and economic recovery.
This dashboard is powered by SafeGraph Places Patterns data, an aggregated, anonymized, privacy-safe summary of foot traffic to millions of POIs in North America. Here, we aggregate the data by categories (like airports or supermarkets) and brands (like Costco or McDonald's). The population sample is a panel of opt-in, anonymized smartphone devices, and is well balanced across USA demographics and geographies.
Academics can get this data for free by joining the SafeGraph Community. We’re proud to partner with thousands of researchers and be a part of the new studies being released every week.
Non-academics can browse our available datasets, and reach out with any questions. We are happy to help.
To build these visualizations we start with a count of visits to individual POIs by day (i.e., the SafeGraph Patterns column 'visits_by_day'). We then further aggregate by summing the data across various industries (e.g., Supermarkets).
The visit data and the underlying panel is dynamic (growing and changing every day). Therefore, instead of showing raw measurement counts, for these graphs, we normalize, standardize, and smooth the data to enable better year-over-year (YoY) and change-over-time analysis.
First, we normalize daily by a measure of the total composition and activity of the panel. Therefore, the y-axis represents a normalized measure of foot traffic happening at this category or brand, given the sample size (which changes day to day).
Second, each year's data is standardized on the y-axis to early March (specifically the median value between the 2nd and 3rd Wednesday of March). This is to better visualize YoY differences and to abstract the semi-arbitrary units of normalized SafeGraph panel visit counts.
Third, we apply a lagging seven-day rolling-window smoothing algorithm. Most categories and businesses have large weekday vs weekend variability. We smooth over seven days to focus the analysis on the overall trend in a category or brand across many weeks, while still capturing important day-by-day changes.
