However, we suspect that the visit patterns may vary across different type of stores. Our hypothesis is that we have changed our shopping behavior during the pandemic. For example, we visit Fast Food and Whole Saler restaurants more often comparing to full service restaurants and typical supermarkets. In particular, we are interested in the following store categories with their NAICS codes:

• Big Box Grocers: 452210 and 452311
• Convenience Stores: 445120
• Drinking Places: 722410
• Full-Service Restaurants: 722511
• Limited-Service Restaurants: 722513
• Pharmacies and Drug Stores: 446110 and 446191
• Snack and Bakeries: 311811 and 722515
• Specialty Food Stores: 445210, 445220, 445230, 445291, 445292, and 445299
• Supermarkets (except Convenience Stores): 445110

• year: column is used for showing the trend line category (orange or blue).
• date: denotes the day of the year for each data point, for which we project to to year 2020. We chose 2020 as the base year because it is a leap year and would have all possile dates (i.e. month + day combination). The actual date for the data point would be month and day from date combined with the year in year.
• median: is used to draw the solid line describing the median visit counts across all stores for that date.
• low: the lower bound of the "confidence interval". In our plot, it is the median minus the standard deviation but will be kept at 0 or above.
• high: the higher bound of the "confidence interval". In our plot, it is the median plus the standard deviation but will be kept at 0 or above.

NOTES

• low and high value will be used to create the transparent area that we see in the plot.
• low, median, high should be computed not only for stores that had visits but also for all stores in Core Places that fit the category. As we learned previously, restaurants with no visits will not be reported in the Weekly Pattern data set.