At a 20% person budget, counts of partners in other groups raised mean next-day cross-group pair coverage by 5.3 points in five SocioPatterns records

We tested four rules for selection of people from one day of SocioPatterns contact data. The target was coverage of contact pairs between supplied groups on the next observed day. A pair was covered when at least one endpoint was selected. The test used five records: two office campaigns, one hospital, one primary school and one high school. There were 26 day transitions. At a budget of the ceiling of 20% of training-day people, the count of partners in other groups gave 52.12% mean coverage. The count of all partners gave 46.79%. The equal-record mean gain was 5.33 percentage points. Four record means were higher; the primary-school change was -0.70 points. Exact expectations handled cutoff score ties. The gain remained positive at 10% and 30% budgets and with contact-time weights. A later control shuffled scores within groups. It gave a 2.81-point gap between rules, compared with 5.33 points in the data. Thus, group selection accounts for part of the advantage under this control. These are descriptive results from short convenience records. The office records share a setting, and daily results are dependent. We do not estimate a population effect or an infection-control benefit. The package preserves code, source hashes, all aggregate outputs, figures, checks and repeat instructions. An independent rebuild matched 1,352 primary output fields. Source data remain external under their own terms.

temporal networks · contact coverage · group structure · person selection · SocioPatterns · reproducibility