Twitch Analysis

This Analysis has the ambition to study the Twitch network. This repository was refactorized to be lighter (with compressed files).

Read Before Using

• For all the files (python and notebook) to run properly you should first unzip the files graphs.zip, Streamers_fr_1D.zip and Streamers_fr_1W.zip see git-lfs for retrieving them.*

• Make a directory Streamer_fr in which the data from the requests will be dumped. If you want another name of directory do not forget to change the variables in the python files.

Gathering the Network Information

Twitch API

Getting the information through the Twitxh API really is the way to go! Yet it suffer from a major disadvantage: you can only get the top 100 streamers for your request.

Twitchtracker

Scraping the awesome Twitchtracker website is simple but not "cool". Since this website tracks Twitch it must possible to do it ourselves, surely by scraping the home page of twitch.tv.

Link Between Channels

• Two channels will be consider linked if a viewer has watched both of them over a given amount of time. The more viewers the more stronger the links.

• The precedent formulation is, in a sense, equivalent to keeping the viewers. Yet keeping the viewers is memory costly.

Results

Visualisation results can be found in the folder ./images, here is one:

Flow Over Time

Another study could be on the time point of view and the flow of viewers. We try to give some insight for channel recommendation at the end of the notebook.

File description

API Request

The file api_req_streams.py is meant to order streamer request. By default it runs for a complete day and store the results in the folder Streamers_fr/.

Twitchtracker Scraping

The file scrape_streams.py is used to scrape the Twitchtracker website.

• ⚠️ modularity is not guaranteed.

Data Gathered

Contains the data of streamers for one given day. See the notebook for more inforamation.

Building the Network

The file build_network.py is used to build the network by default from the folder Streamers_fr/. It can be slightly modified to keep the viewers. It mainly use the librairy networkx.

• The variant serve different purposes but are alike.

Network

Graph used for the notebook and obtain from the data above.

Analysis Notebook

The file twitch_analysis.ipynb is a notebook to analyse the Twitch network/data.

Requirement.txt

Python librairies requirements

Visualisation Gephi File

Images of representation obtained with Gephi.

• ⚠️ it correspond to the filtered graphs (see the notebook).