TwitchFlare

TwitchFlare is a visual representation of the top games on the Twitch.tv platform. The app gathers data every hour, and displays it within a Sunburst Chart, which allows users to see trends between not only the top games throughout the day, but also streamers that are competing with each other.

Approach

TwitchFlare only visualizes the past 24 hours worth of data, each hour at a time. This is due to restrictions on storing data from Twitch, as stated in their Terms of Service Article 12 Section b. With that said, I focused on three main points for this project:

1) Retrieving and Storing Data

Via the Twitch API, I was able to get the most current data about the games their streamers show. Unfortunately, the API has multiple endpoints for different pieces of data. I used ajax calls to get the data for the top games at that point in time, and then. This only gave me data on the specific games though, so I made a separate ajax call for each of the games to find its top 10 streamers with the most views, and then store the relevant information in Mongo as a single document.

D3js requires data to be formatted in a specific manner for each type of chart, so I spent just as much time on parsing my data properly as I did on making my charts visually attractive. This is all happening in the background on the server, thanks to Node-Schedule, which allowed me to set up a CRON Job, or a recurring task.

2) Visualizing Data from the Past 24 Hours

Using a Sunburst chart, I was able to show my data, not only by count, but by size, or number of viewers. Since I didn't want to pull in data for 400+ games and 1000+ streams for each of the top games, I limited my API calls to only 6 games, and 10 streams. The data I pulled only shows the top 10 streams ranked by viewership.

3) Making the Data Interactive and Visually Appealing

With only the size and the count settings, I thought that the chart gets cluttered, especially for the games with less viewership, such as ArmA II and Destiny. Games like League of Legends, Dota 2, and Counter-Strike: Global Offensive will always have higher viewership since they're more popular. Thus, I added the ability to zoom in on each entry in the chart.

You can zoom in by clicking on a section, or zoom out by clicking on the center of the chart. I also added in a tooltip window that automatically updates with the stream or game's information and logo.

Technologies Used

• HTML
• CSS
• JavaScript
• jQuery
• MongoDB
• Express
• Node.js
• Node-Schedule
• Twitch.tv's API
• D3js

Notes

Author: Jonathan Lam Last Updated: 10/16/15

Plans for Future Versions

• Ensuring bad data doesn't enter the database
• Ensuring bad data doesn't break the chart
• Dynamically updating a table at the bottom, updating with data about that specific game