This chapter explores the role time has to play in social systems and tools to help you understand that role. You start by looking at models of how you intuitively think of and form assumptions about the timing of social media events. Then you revise these assumptions by observing that the temporal characteristics of actions differ considerably from what you expected on data sets of Tweets and Wikipedia posts. As in the previous chapters, these observations may seem counterintuitive at first but hint at the presence of large variances along the temporal dimension.
The way we live our lives is strongly determined by the cyclical flow of time on many different levels. The repetition of days, weeks, seasons, and years suggests that there have to be similar patterns observable among the events in social media as well. We identify these trends in our data as well and present a framework on how you can use these to your advantage and forecast future metrics in time.
In any dynamic system time plays a role. A social system has many events in time: Users join the network, add edges, click links, make searches, and send messages. All these events have a time associated with them. This chapter shows how to analyze streams of events in time quantitatively.
The most complete way to represent a series of events is as a sequence of timestamp–event data pairs, where the event data ...