Network

We build a network using actors who starred more than 25 movies in the TMDb database as nodes. It turns out that there are 38 actors match this standard. And co-starring relationships between every two of them are edges. There are 240 edges in the network.

Basic facts

The followings are some basic facts of the network:
Average degree: 12.6316
Graph diameter: 3
Graph density: 0.341394
Number of connected components: 1

Analysis

As we can see, nodes in the network are closely connected to each other, for the diameter is rather small. It indicates that these movie actors often cooperate with each other.

We conducted community detection on the network by Louvian method which maximizes modularity. There are four communities in the result, which contains, in decreasing order, 13, 12, 10, and 3 nodes. The largest community (nodes in green) contains actors starring in more than one super hero movies of Marvel or DC Comics. Though not all of them play leading roles, many of them have multiple participations as supporting roles. The second largest community (nodes in blue) contains many actors who have played sort of 'tough guys' in action movies, e.g. Channing Tatum, Liam Neeson, Matt Damon or John Goodman. Also actors like Morgan Freeman, who does not play that kind of 'tough guy' but played many times as their close friends, partner, mentor or something, were clustered into the same community with 'tough guy' actors.

Notice that actors with high reputation on their performance skills do not appear in this network. It might indicate that some actors make money by quantity of their works (who are in the network and also quite famous) and some make money by quality of their works rather than quantity.

This analysis also indicates that it might be helpful if we treat the cast (or crew) of a movie as a features for predicting box office performance. In our project we have already showed that the number of cast or crew can be a indicator of performance. It is our guess that the exact combination of casts (or crew) can further imporve the results of predictions.