Methodology: Here, we want to assess how the contact network changes by removing some of the time intervals for which we had a peak of contacts. We do so because there are discussions for implementing policies that allow people to move freely on specific hours within the day. To analyze this behavior, we build all the 5-minute interval graphs per day, and then we remove the top N 5-minute intervals in terms of the number of contacts that occurred. Then we report the average number of contacts per person, the number of contacts per day, and the percentage of contacts that remain after applying each policy. Below, we present our results and the main insights by analyzing this behavior for varying numbers of N.

Main Insights: Our experimental results indicate that a large percentage of the contacts per day happen within a few hours. For instance, by looking at the last column of Table 1, we observe that by removing the top 1 hour we remove almost 13% of all contacts and by removing the top 2 hours we remove almost 24% of all contacts. In this setting, if we aim to cut the contacts into half, removing the top 5 hours per day seem to be an effective strategy. We also note that the overwhelming majority of the contacts per day are happening within the top 12-hours (only approximately 17% of the contacts remain). By looking at the temporal plots in Figure 2, we observe that these policies are particularly effective during the weekdays and less during weekends. This is likely because on weekends the numbers of contacts are more spread out with time, while on weekdays most of the contacts are concentrated in a few hours, mainly because of courses, social events, etc.