a, b, The POIs from all metro areas are pooled and the quantities from the mobility data are shown. a, The distribution of dwell time. b, The average number of hourly visitors divided by the area of the POI in square feet. Each point represents one POI; boxes depict the interquartile range across POIs; data points outside the range are shown as individual dots. c, d, The data are pooled across model realizations from all metro areas and model predictions are shown for the increase in infections (per 100,000 population) because of reopening a POI category. c, Data per POI. d, Data for the category as a whole. Each point represents a model realization; boxes depict the interquartile range across realizations; data points outside the range are shown as individual dots. Across MSAs, we model 552,758 POIs in total, and we sample 97 parameters and 30 stochastic realizations (n = 2,910); see Supplementary Table 6 for the number of sets per metro area. Colours are used to distinguish the different POI categories, but do not have any additional meaning.