Methodology:
The goals of the methodology applied were twofold: one, to explore what fractions of users can be expected to be present in the network by what time of day, and visually interpret the resulting visualizations; and to determine whether there was variation between weekdays as well as between weekday and weekend in Lower Manhattan client duration and first seen times.
To this end, a cumulative probability function for client device presence in the network on weekdays as well as weekends was plotted (Figure 3, 4). To address whether specific days of week were affecting the duration of users in Lower Manhattan in the network, or whether there was significant variation in the times workers arrive in the network each day, two-sample Kolmogorov-Smirnov (KS) testing was performed on distributions of user duration and arrival times in the network. KS testing is useful in this context because of the complex and exploratory nature of the data; a KS test does not assume that the data come from any predefined distribution, and instead can be used to test the similarity of two sets of data.