For this exam you are free to use any course or external materials, but you are not to give or receive answers to this exam. You may not talk to anyone about the exam before you (and they) have finished it.
Once you start the exam, you have 2 hours to complete it, and you have only 1 attempt.
Good luck!
Question 1
The following 4 visualizations depict the same network. Which one follows the guidelines for effective network visualization?
Question 2
Which of the following networks has the highest clustering coefficient?
Question 3
The maximum number of possible edges in an undirected network of 9 nodes is
Question 4
The degree distribution corresponding to the network visualized here is:
Question 5
Zane has 4 friends, Amy, Bob, Clive, and Diane. Of his friends, Bob and Clive know each other but none of his other friends do. What is Zane's clustering coefficient
Question 6
The probability of any two nodes being connected by an edge is equal. This description fits the
Question 7
Consider two networks N1 and N2, with the same number of nodes and edges. The degree centralization of network N2 is much higher than that of N1. Which one is more likely to be susceptible to targeted attack where a certain number of nodes are selected and removed?
Question 8
Assume that initially one node in the network is infected with a virus. At each time step each infected node infects each neighbor with some fixed probability. On average, in which of these two networks will the virus propagate more quickly?
Question 9
In the following network, which node has highest betweenness.
Question 10
In the following network consisting of just 3 nodes, the betweenness of A is:
Question 11
Assume nodes are innovating and communicating over a small world topology. They are navigating Kauffman's NK fitness landscape in trying to achieve the best solution. That is, each of them is trying to figure out a sequence of 0's and 1's that has the best score. But there are many local maxima (pretty good solutions), where flipping any individual bit will produce a worse solution. Each node will look to its network neighbors, and copy the best solution they have, if it is better than their own. If no neighbor has a better solution than the node's current solution, then it will try to flip one bit in its current sequence at random and keep the change if it has a superior fitness to its current solution. Which of the following is true about such a process on a small-world topology?
Question 12
Consider an SI model (nodes are susceptible until they become infected at which point they remain infected), where two network neighbors are initially infected. Infected individuals are designated with yellow nodes, non-infected individuals are designated as white nodes. The following image could have been produced by which of the following forms of contagion:
Question 13
Which of the following partitions of nodes in the same network into communities will produce the highest modularity?
Question 14
Which of the following networks is the most disassortative on degree (has negative degree assortativity).
Question 15
Two nodes, A and B, have equal indegree but the PageRank of A is higher than the PageRank of B. A possible explanation is that:
Question 16
Among these actions, which one (if applied repeatedly) is most likely to reveal community structure present in a network?
Question 17
In the small world network model, if the probability of random rewiring is set to 1, you obtain what kind of network?
Question 18
How many strongly connected components are in this network?
Question 19
If the probability of passing a message on in a small world experiment is 0.5, no matter the position the person occupies in the chain (e.g. 1st, 2nd, 3rd, ... last), then the probability of a chain of length (n + 2) getting through relative to the probability of length n getting through is:
Question 20
You developed a model of network formation. You are checking the properties of the networks your model produces against an empirically observed network you are trying to model. You discover that the motif profile of the model-generated networks is different from the empirical network. What can you conclude?