October chatter data from vrabi (blue line) for all holoEN streams/premieres recorded. Anything less than 20* minutes was not included (revision). Stream-time was not considered (longer streams correlate with more unique chatters, usually). Max and min apply for final averaged amounts. The other three are: average of Top 50% streams sampled, Average of all streams sampled, and average of Bottom 50% streams sampled. Average viewership might be cool to consider if the data set didn't require more manual work and a bunch of assumptions.
November 2024 here:
>>91310468>DISCLAIMERS - Chatters are only a proxy, non-EN chatters are less inclined to chat. Stream content also affects chat participation.
- Gura and Ame both had one stream. Data was excluded from graph since both would render the graphs meaningless
- Chart was expanded to top 5 and bottom 5 on account of Gura and Ame
- Notes in the chart is there to add context
>Categories Notes:- StarsJP, NijiJP, and VSpo! help indicate how JP the talents chat is. Below 2-3% might not be statistically significant though. HoloJP and DEV_IS technically do too, but I'm not quite sure how to parse the difference in chatters between the two/
- StarsEN, NijiEN, and Phase indicate how EN the talent's chat is. Same spiel about statistical significance.
- The more NijiJP>10% HoloJP is true, the more "JP" chatters the talent has.
- The StarsJP:StarsEN ratio is pretty important too. The closer the ratio is to 1:1, the more JP chat is. This holds true for the NijiJP:NijiEN ratio as well.
>Observations:- Comparing the girls with one another really doesn't work because of total unique chatter size, stream-length, average CCV, etc. Probably need a factor to "normalize" things. Would require a bit of testing, but bweh.
>For shitposting:-
If Phase>Stars EN, then the girl is more CGDCT-codedWill be doing December. A September one would be cool to see if a graduation "buff" can be id'd. Would require me doing August though, so there'd be at least a sample of two. A lot of manual data entry though...
