1. Time series analysis
We’re very close to the finish line! Just one more set of exercises and we’ll be done. In the following exercise set, you’ll be performing
2. Time series
time series characterization of a particular set of nodes in the graph. Do you recall the distinction between “global” and “local” analyses? Global generally referred to how the distribution of some statistic over all nodes changes as a function of time; local generally referred to how the statistic of one (or a small group of nodes) changes over time. In this set of coming exercises, you’ll be building a picture of how degree changes over time for a small group of nodes. Let’s now quickly recap some code you’ll need to build this rich and complex picture of local connectivity!
3. Datetime arithmetic
Firstly, you'll be using the timedelta object. A timedelta object expresses a time difference between two date or datetime objects. To express a time delta of 4 days, we can create a timedelta object td, where we pass in days as a positional argument. (The first position in the timedelta constructor is the number of days.) Once we have that, we can do date arithmetic by doing, for example, date1 + td. Next up, let’s quickly recap
4. Degree centrality
degree centrality. Again, this concept should be familiar by now, so I will use a toy graph as the motivating example here. Given a graph G, to obtain the regular degree centrality metric, you only need to pass in the graph, and a dictionary will be returned with the keys as the nodes and the values as the degree centrality metric. To obtain the bipartite degree centrality metric, the nodes that belong to one of the partitions has to be passed in. This still returns the degree centrality of every node, but note the values will be different from regular degree centrality: do you remember why? Pause the video for a moment… if you said “the denominator changed”, you’re right! The denominator is the number of nodes that any node could possibly be connected to; in the bipartite case, it is only the number of nodes on the opposite partition.
5. Let's practice!
In the exercises that follow, you will characterize the time-dynamic popularity of forums in the network dataset, and use that to find out which forums were the most popular over time. Let’s go and tackle the exercises!