- 5 min read. Content may take a few seconds to load.
Mobile Device Visits to a Single Full-Service Restaurant
Remote sensing is the art and science of acquiring information about something without making direct contact with it. Think of mobile data like high-resolution remote sensing of the market for an important industry in Hawaii like accommodation and food services. When a mobile device dwells at a food service point of interest for at least four minutes, which in the SafeGraph panel sample is the minimum for what constitutes a “visit,” it is like a remote sensed record of the fact that someone probably made a buying decision. The machine learning problem for SafeGraph, which they detail in an interesting white paper, is how do you ground truth where a person was actually going based on where their mobile device appears to dwell?
Mobile Device Visits to a Single Full-Service Restaurant Showing the Impact of COVID-19
The Data: The star of this data dish is a single full-service restaurant point of interest (POI), Ken’s House of Pancakes in Hilo. The graphs below shows raw counts of mobile device visits to Ken’s House of Pancakes from the SafeGraph panel of mobile devices (raw_visit_counts).
What To Look For: The graph shows an overview of visits to Ken’s House of Pancakes before, during, and after the COVID-19 public health emergency compared with mobile device visits to all full-service restaurants statewide. You can see how its recovery was especially impacted by a closure in late August 2020 after one employee tested positive for COVID-19. We imagine no one was more upset about Ken’s being closed, again, than Dwayne Johnson.
Scroll over the graph below.
Home Location of Mobile Device Visitors Pre and Post-COVID-19
The Data: The map below shows where visitors to Ken’s House of Pancakes came from by census block group based on their “home” location (visitor_home_cbgs). The “home” location of a device is determined by analyzing the last six weeks of dwell times during nighttime hours from 6 pm to 7 am local time.
What To Look For: Look at the largest circles, as well as the full extent of the “home” locations of mobile devices that visited Ken’s House of Pancakes. The size of the symbol is proportional to the number of mobile devices that visited Ken’s House of Pancakes. In both maps below, use the reset view button 
to return to the original extent. The zoom to POI button 
will zoom in on the POI’s location. When you mouse over an individual census block group you are presented with the following details. For example:
8 visitors | 102 devices | 1536 (+- 262) est. pop. | 1.3 HI | 0.9 US
KEY
No. of visitors to Ken’s from CBG | No. of devices in SafeGraph panel from CBG | Estimated population of CBG (American Community Survey 2016) | Index of devices per capita in CBG as compared to the State of Hawaii | Index of devices per capita as compared to United States as a whole. Index > 1 means there are more mobile devices represented per person in this CBG than the rest of the state or the United States as a whole.
Zoom in, turn on layers, and scroll over the map below.
Daytime Location of Mobile Device Visitors Pre and Post-COVID-19
The Data: The proportional symbol map below shows where visitors to Ken’s House of Pancakes came from by census block group based on their daytime location (visitor_daytime_cbgs). Daytime location is determined by looking at 45 days of data about where the device dwelled most frequently during daytime hours between 9 am and 5 pm local time.
What To Look For: Like the map above, look at the largest circles, as well as the full extent of the “daytime” locations of mobile devices that visited Ken’s House of Pancakes. The size of the symbol is proportional to the number of mobile devices that visited Ken’s House of Pancakes.
Zoom in, turn on layers, and scroll over the map below.
Popularity by Day Pre and Post-COVID-19
The Data: The bar chart below shows data on the number of visits in local time on each day of the week (popularity_by_day) for a single full-service restaurant, Ken’s House of Pancakes in Hilo.
What to Look For: These small multiple bar charts, and similar information products for other data below, bring out subtle changes in monthly, weekly, daily, and even hourly visit patterns to Ken’s House of Pancakes before and after the COVID-19 public health emergency.
See bar charts below.
Popularity by Hour Pre and Post-COVID-19
The Data: The bar chart below is data showing the total number of visits to Ken’s House of Pancakes in each hour over the course of the date range in local time (popularity_by_hour).
See bar charts below.
Bucketed Dwell Times Pre and Post-COVID-19
The Data: The bar chart below is data on the distribution of visits by dwell times based on specified categories of dwell time (bucketed_dwell_times). The categories are dwell times in minutes and the values are the number of visits in each dwell time.
What To Look For: For any given full-service restaurant, or for all full-service restaurants across the State of Hawaii, one of the indicators of recovery might be a return to the same basic distribution of dwell times as before the COVID-19 public health emergency. For example, for the State of Hawaii average, the two most popular dwell times were 21 and 60 minutes, followed by 5 to 10 minutes. During the COVID-19 public health emergency, when visits were significantly lower, the most popular dwell time shifted to between 5 and 10 minutes, likely indicating a change in behavior related to more take-out orders.
See bar charts below.
Map of Related Same Day & Month Brands
The Data: The proportional symbol web map below is data about the top brand name points of interest that the visitors to Ken’s House of Pancakes also visited either on the same day (related_same_day_brand) or in the same month (related_same_month_brand). The example below uses the maximum value for related same day or same visits for all months in 2018.
What To Look For: Look at the largest circles. The size of the symbol is proportional to the percent of visitors to Ken’s House of Pancakes that also visited the brand on the same day or month, and the color of the symbol is explained in the legend. For example, in 2018 (“Related Same Month Brands 2018”) nearly half (44%) of the mobile devices that visited Ken’s House of Pancakes also visited the nearby Walmart in the same month. Likewise, 13% of the mobile devices that visited Ken’s House of Pancakes also visited the nearby Texaco on the same day.
Zoom in, turn on layers, and scroll over the map below.
Bar Chart of Related Same Day & Month Brands
The Data: The bar chart below shows the top brand name points of interest that the visitors to Ken’s House of Pancakes in Hilo also visited either on the same day (related_same_day_brand) or in the same month (related_same_month_brand).
What To Look For: The largest bars in the bar chart are the most closely related brand locations to Ken’s House of Pancakes.
See bar charts below.
TECHNICAL NOTES
SafeGraph variable names are in parenthesis. For detailed information on how SafeGraph variables were derived, see the SafeGraph documentation. SafeGraph data was processed for the State of Hawaii using PostgreSQL and PostGIS then added into R. SafeGraph point of interest geometry data, including highly-accurate building footprints, is being continually updated. SafeGraph strives for coordinate accuracy less than 10 meters away from an accepted coordinate truth set (Google Maps). Please let us know if you recognize any coordinate accuracy issues in the web maps presented above. Graphs and bar charts developed in R included use of the dygraphs and ggplot2 packages. Web maps developed in R included use of the leaflet package.
