Characters Network

About the Network

From the 313 characters information we just talked about, we created a "Characters Network".
- We used both Rickipedia and the Rick & Morty API to come up with the numbers above. We took the well-formatted information about characters from the API and we crossed them with the contents of the wikipedia pages. To create the network we parsed each Wikipedia page of the characters to find hyperlinks to other characters. When a link was found we created an edge (a link/connection) from the given character to the other character.

General View

Below is a simple visualisation of the network that we generated. If you haven't watched the show, you must know that it contains a lot of duplicate / alternative versions of the two main characters, Rick and Morty.

Above you can see the complete character network, we highlighted the Ricks in blue to match Rick's pointy, mad scientist hair and the Mortys in yellow to match the bright yellow t-shirt that he wears throughout the show. We also scaled the node size depending on their degree (how many links/connections they have) which is why Rick & Morty have such big nodes. The three black nodes that are also fairly big represent the remaining members of the Smith family - the father Jerry, the mother Beth and Morty's bigger sister Summer.
Now - Let's dive a little bit deeper into the analysis of our network, by looking at how the characters are connected within the network.

Degree Analysis

In the figure below, we plotted the degree distribution, i.e. number of nodes with a given number of links. On the horizontal x-axis is the degree (number of connections from a given node to other nodes), and on the vertical y-axis is the number of nodes with that given degree value. This plot provides insight about how the connections between characters are distributed in the network, i.e. how many characters have few connections to other characters and how many have a lot of those connections.

When we think about it, the plot above shows nothing very surprising. It just confirms something that was expected : Rick & Morty, as well as their family members, have a lot of connections compared to the remaining characters. What is more interesting is that lots of other characters only know a few others, which makes sense considering the structure of the series. As the show focuses on entirely different random adventures for each episode, it's logicial that only the main characters living those adventures have a lot of connections to a lot of other characters whereas most of remaining characters will never get to meet each other.

Centrality

Centrality assigns an importance score to a node based purely on its number of connections. So by taking a look at centrality, we can get a glance of which characters are the most popular, or which character to know if I want to know another one. So let's look at which of our characters are the most central.

Betweenness Centrality

As expected, Rick and Morty are most central, followed my the remaining members of the Smith family. Jerry is probably third because he joins Rick and Morty on most adventures, compared to the other two family members. So if I'm a new character in Rick & Morty and I want to meet another one, the optimal way is to ask Rick first, then Morty, then Jerry (you might not want to do that, or the conversation might get very boring), etc.

Centrality Eigenvector

The Eigenvector centrality is a measure of how nodes (characters) are connected to other nodes with high degree (many links/connections), and in that sense it is also a ranking. Characters have high eigenvector centrality probably because of popularity, i.e. they appear often with other high-degree characters.

Again, Morty and Rick appear with highest centrality, whereas Jerry is on fifth place. Perhaps, like explained above with betweenness, because he is on many adventures in different locations, where a lot of characters with low degrees appear. So that makes Jerry popular...

Everything as expected

Well so far we haven't realized anything surprising. But is it a bad sign ? No it's actually a good sign because it shows that the data we collected is in correlation with the show which is a very good things this far!
Rick wins the betweenness centrality contest and Morty has the highest eigenvector centrality. These results match our expectations very well, and could most likely be explained by the fact that Rick knows almost each and every character across all locations and episodes, therefore he has the highest betweenness. Morty on the other hand, is a central member of both the adventures and the Smith family, which makes him more connected to other nodes with high degree than Rick.

Status View

Another fact about Rick & Morty is that it involves LOTS and LOTS of death. Luckily our data sources included this information about the characters so we decided to look into that, to find the "most dangerous character" of the show. We started by plotting the following network, depicting alive characters as green nodes, dead characters as red nodes, and those nodes with "unknown" status as black :

Mmmh out so out of 313 characters 114 are dead... Can we find the responsible for all those dead bodies, who could be the most dangerous characters of the show?

Most Dangerous character

We ran a little analysis about this through our Network. We looked at every nodes' neighbors (the nodes that are just 1 link/connection away) and analyzed their status (alive or dead) to have a sense of who has been in contact with the biggest amount of dead characters.

And again, it's without big surprise that Rick & Morty are the most dangerous characters. But the otherwise innocent father Jerry Smith gets the third place in our rankings of most dangerous character in the whole series, beating his own wife & daughter. Since these characters have a lot of edges (links/connections) and appear more often, their chance of being next to dead characters increase.

But when we look beyond the regular Smith family we stumble upon the infamous Evil Morty who probably has more kills than we know at that point of the series. Not surprising again but Beth & Summer better work on their danger game during season 4 otherwise Evil Morty might pass them !

Species View

Rick & Morty is a very imaginary, explorative and diverse show, in terms of characters. Especially the variety of species is interesting, and therefore we decided to also plot a network with information about the species of each node.

As shown by the figure, there are actually 10 known species at the moment. We wonder how many more will be added in season 4 on the new adventures of Rick & Morty to distants dimensions. While humans are still the most represented, we really enjoy the presence of all the others !

Communities

This section analyzes the communities within the characters network by comparing four different algorithms for community detection. Plots are used to show the structure of communities and which of the main characters are central to a specific community. For instance if Jerry Smith is the central character in a given community, all nodes within that community will be colored darkgreen to match Jerry's awesome shirt. Likewise for Rick, where nodes are colored turquoise to match his pointy hair. If Rick and Morty are both in the same community, nodes will be colored in a flashy green to recall the portal gun.
For each plot, node sizes are dependent on the size of the communities, such that bigger communities have bigger nodes.

But first, a little note on what modularity means, since we use this number to determine how good a given algorithm is to find communities in a network; Modularity: A number between 0 and 1 that measures how good a division of a network into communities is, where 1 is perfect division.

Greedy Communities

The greedy modularity maximization divides characters into communities by trying to optimize the modularity value.

Modularity : 0.48

As shown by the graph this actually results in community size decreasing as the communities are formed. Since communities are based upon maximization of modularity, a few big communities exist. This is due to the high degrees (many links/connections) of Rick and Morty. This algorithm, as described by it's name, will divide communities with the highest modularity, since that is inherent in the functionality of the algorithm.

Fluid Communities

Modularity : 0.41

The fluid communities algorithm requires defining a number of communities, since the first step of the algorithm is to assign communities randomly to this amount of nodes within the network. Then the algorithm divides communitites by randomly running through nodes and assigning them to communities, based on which communities the neighbors of the given node are divided into. This randomization results in more distinct communities than some of the other algorithms. It is worth noting that changing the predefined number of communities has a high impact on the modularity. The number 15 was chosen, since two of the other algorithms divided the characters into exactly 15 distinct communities automatically.

Label Propagation Communities

Modularity : 0.24

The label propagation algorithm performs worst of all the algorithms, since it has the lowest modularity value. All members of the Smith family are clustered together into one large community that dominates the graph, also visually. This large community exists because of the labelling approach used by the algorithm, where nodes are divided into communities if their neighbors contain similar labels. This approach will enhance communities with high-degree (many links/connections) nodes, thus generating this dominant community containing all members of the family.

Louvain Communities

Modularity : 0.45

What is interesting about the Louvain algorithm is that the approach it uses to divide communities is similar to greedy modularity maximization. The major difference is the initial grouping of nodes locally. Perhaps this local grouping is the reason why the number of communities is lower than the greedy modularity maximization algorithm.

Community Wrap-up

The four different community algorithms divide the characters in slightly different ways. For example, Rick and Morty are not always grouped into the same community. But perhaps more interestingly, Summer Smith is the only member of the Smith family to reside in her own community for most of the algorithms. This could indicate that she is detached from Rick & Morty or detached from the Smith family. Not necessarily jumping to conclusions here, but her not always nice attitude towards Rick & Morty could perhaps explain some of this detachment.
Even though the algorithms provide different results, there are also a number of interesting similarities. For instance the characters Terry and Purple Morty are central within a community for all four algorithms, making them characters worth noting, since they apparently have some central position in the show. All algorithms divide the characters into 15 or less communities. This is quite interesting, since the show contains far many more distinct locations, and we expected a larger number of communities that would be closer to the 100+ locations.