1. Learn
    2. /
  2. Courses
    3. /
  3. Intermediate Network Analysis in Python
  • 1

    Bipartite graphs & product recommendation systems

    Free

    In this chapter, you will learn about bipartite graphs and how they are used in recommendation systems. You will explore the GitHub dataset from the previous course, this time analyzing the underlying bipartite graph that was used to create the graph that you used earlier. Finally, you will get a chance to build the basic components of a recommendation system using the GitHub data!

  • 2

    Graph projections

    In this chapter, you will use a famous American Revolution dataset to dive deeper into exploration of bipartite graphs. Here, you will learn how to create the unipartite projection of a bipartite graph, a very useful method for simplifying a complex network for further analysis. Additionally, you will learn how to use matrices to manipulate and analyze graphs - with many computing routines optimized for matrices, you'll be able to analyze many large graphs quickly and efficiently!

  • 3

    Comparing graphs & time-dynamic graphs

    In this chapter, you will delve into the fundamental ways that you can analyze graphs that change over time. You will explore a dataset describing messaging frequency between students, and learn how to visualize important evolving graph statistics.

  • 4

    Tying it up!

    In this chapter, you will apply everything you've learned in the previous three chapters to a forum posting dataset. You will analyze the temporal changes in forum user connectivity patterns, and make visualizations of evolving graph statistics over time.

Initializing

Exercise

Plot number of posts being made over time

Let's recap how you can plot evolving graph statistics from the graph data. First off, you will use the graph data to quantify the number of edges that show up within a chunking time window of td days, which is 2 days in the exercise below.

The datetime variables dayone and lastday have been provided for you.

Instructions

100 XP
  • Define a timedelta of 2 days using the timedelta() function and specifying an argument for the days parameter.
  • Inside the while loop:
    • Filter edges such that they are within the sliding time window. Use a list comprehension to do this, where the output expression is (u, v, d), the iterable is G.edges(data=True), and there are two conditions: if d['date'] is >= curr_day and < than curr_day + td.
    • Append the number of edges (use the len() function to help you calculate this) to n_posts.
    • Increment curr_day by the time delta td.
  • Make a plot of n_posts using plt.plot().