Search by Category

Interview Question: Friend Circles

There are N students in a class. Some of them are friends, while some are not. Their friendship is transitive in nature, i.e., if A is friend of B and B is friend of C, then A is also friend of C. A friend circle is a group of students who are directly or indirectly friends. You have to complete a function int friendCircles(char[][] friends) which returns the number of friend circles in the class. Its argument, friends, is a NxN matrix which consists of characters 'Y' or 'N'. If friends[i][j] = 'Y', then i and j students are friends with each other, otherwise not. You have to return the total number of friend circles in the class. Note: The method signature will differ by language. For example, Java will have 'int friendCircles﴾String[] friends﴿' where "friends" is an array of strings, which can be viewed as a 2 dimensional array of characters.

Constraints:
- 1 <= N <= 300
- Each element of matrix friends will be 'Y' or 'N'
- Number of rows and columns will be equal in friends
- friends[i][i] = 'Y', where 0 <= i < N
- friends[i][j] = friends[j][i], where 0 <= i < j < N

Question Format

Online coding challenge

Time Constraint

1 of 3 Coding problems 2 Hours given

Context

This question was a 2nd round automated interview question given after passing multiple choice statistics question

Initial thoughts

When I saw this question I made a critical error in my initial understanding of what a friends circle meant.

I was able to make the first key insight which is that the friends and relationships are a structure called Network Graph. The problem is asking to find the number of unique relationships that span friends. I first ran into my first issue here.

Because of the title Friends Circles I became fixated on the thought that a closed loop is what I was looking for. Loops in graphs are called cycles and numerous algorithms exist to find them. However this is not what this question was asked for and I wasted pretty much 40 minutes implementing this solution.

Where I went wrong

Not understanding the question

During the interview I wasn't able to understand the question correctly. Unfortunately since I had no human interviewer I was unable to ask for clarification

Too slow to iterate

It became apparent to me that my understanding was wrong when I failed the majority of the test cases. But I had already wasted 30 minutes to get to that point and ran out of time

The Solution

It was misleading that the problem was titled Friends Circles because the problem is actually asking how many people are connected through their friends and their friends. In math terms it's asking how many groups of nodes are connected by their edges.

Given this understanding the problem is extremely simple. Any graph can be composed into a tree. This is a concept called tree composition. In this problem all we needed to do was pick a node, run a depth first search algorithm to find all the nodes that share a tree, then move onto the next node and try again. The number of unique trees is the answer to all the test cases.

Code

The first order of business is to read in the inputs that look like this

YYNN  
YYYN  
NYYN  
NNNY  

into a sparse matrix that looks like this

network = {0:[1]
           1:[0,2],
           2: [1],
           3:[]
           }

I chose the sparse matrix because beside being very human readable it also lends itself to the search process.

Given a sparse matrix we can now find circles by doing a depth first search of that person's friends as shown below.

def find_circles(friends_dict):
    '''
    Iterate through friends dict to find 
    '''

    def find_friends(person):
        '''
        Recursive function that does Depth First Search of all friends
        '''
        if person not in visited:
            visited.append(person)
            for friend in friends_dict[person]:
                find_friends(friend)
        return

    visited = []
    circle = 0
    for person in friends_dict:
        if person not in visited:
            find_friends(person)
            circle +=1

    return circle

The algorithm works initially picking a node. It then traverses the tree to find all others nodes (friends) that are connected with the initial node. When it exhausts all friends in the chain it tries the next node.

In particular is the find_friends function. This is the workhouse of this implementation. It recursively finds the friends of the current node. The find_circles function mostly acts as an outer wrapper, storing the friends count and the visited list.

The visited list is particularly important because it ensures two things. One is that we don't waste computational time finding all the friends of a person we've already searched. Additionally, it allows the find_friends function to terminate once all the visited friends in a tree have been found.

Visualization

Here's a visualization of the search process. You notice that once the algorithm finds a tree it stays on it until there are no more nodes left.

To make this visualization the algorithm was also reimplemented in javascript which you can find here. The visualization is possible due to the awesome d3 javascript library which will get a post of its own in the future