Visualizations.txt

Questions Tim suggested and some comments:
How successful are commercial repositories compared to private repositories -> Don't know how to determine if the repo is commercial?

How to make a GitHub repository more successful -> Maybe try something without a model? Can we see some trends of popular contributors connecting
popular GitHub repos? Is that a reliable metric? That would run in parallel to how one can make it in the industry (-> If you know somebody who already made it big)
-> Look at the same creation year, same topic/keyword -> Very tough question, but high impact

General purpose system that allows general exploration -> Need to argue why our system is good/what questions it can answer.

-----------------------------------------------

How do we evaluate our visualization platform
-> formulate some questions by ourselves
-> build a system that can answer most if not all of them 
-> We also need to explain the whole process, so that they know how challenging the project 

-> Include that we had to pivot our project right before the first check in and why
-> What were the challenges with the new approach, scraping etc., why did we choose the visualization we did
-> 10-15min presentation


-----------------------------------------------
Brainstorming:

Question to answer:
- Does a successful contributor automatically make the repository successful (what's the metric of success)
- Can we group related topics together efficiently -> have a multilevel graph, KNN style
- Size of node: Popularity of Repo, width of edges: Popularity of contributors
  -> Do popular contributors only work on popular repos
  -> Filter the displayed repositors by number of popular contributors and then display them

- Is GitHub's keyword assignement accurate w.r.t. connectivity of the repos/topics
- Repo to repo, color by keyword -> are there distinct groups of colors, or mixed


Visualization ideas:
- topic to topic dataframe with # of shared repositories -> first and easiest one
- topic to topic dataframe with # of shared contributors -> slider to change the threshold of needed shared contributors -> How do the connections look like

color coding topic bubbles by language used (%) -> slider to change creation date of the repos
--> See changes in language distribution -> Can answer the question of how fast languages spread in the coding world & what their impact is?


Evaluation:
- Check if the questions can be answered and maybe include the graphs for each of the questions 
- If time: Answer the questions with traditional visualizations (for example Emma's visualization of Jupyter Notebook's spread)
  --> Compare the two, does the interactive network make sense?