TwitchOverlap VS np-overlap-data

The problem is there is too many edges in the graph to really examine more intricate details without loading the graph in a tool like gephi or writing something custom to visualize the data. Making the lines thicker makes the entire thing even more unreadable. Filtering out some of the less weighted edges might help but I'm not sure. I was hoping there would be cleaner separation between communities like in the more general twitch overlap graphs. Using the sample data provided at https://github.com/snoww/TwitchOverlap/tree/master/data and running the same algorithms in gephi I was able to get communities to cleanly separate. I was unable to get this to occur with my dataset. I theorize that this is due to far greater overlap in the nopixel gtarp community than normal as well as the fact that I did not attempt to filter out bots or people using apps like chatterino which would cause them to show in a ton of chats at once. This is reinforced by the fact that the sample data has an average degree of 16 whereas this graph has an average degree of 115. In addition, many of the streamers in the community are far smaller and therefore additional filters may remove them entirely from the graph. For anyone interested in taking a closer look at the data I would recommend downloading the raw data then following the instructions here https://github.com/KiranGershenfeld/VisualizingTwitchCommunities#visualization-tutorial to load the data in gephi where you will be able to run algorithms and take a more detailed look at how the different communities interact with each other.

2022-01-nodes.csv

Someone made new automated Twitch atlas graph and posted it to LSF here. Also here is the pic from LSF. https://i.redd.it/bw414g8k3h571.png. Data yoinked today and it was updated last yesterday in his github page. https://github.com/snoww/TwitchOverlap

The graph shows the overlap of viewers in the top 1000 channels (by unique viewership) in the month of June, and then using the Louvain algorithm for community detection within the graph. The data is collected hourly automatically using Twitch's api, a detailed explication of the data collection and processing can be found on the GitHub

There’s a graph not destroyed by Twitter that’s easier to see on the actual GitHub page