[ 3 / biz / cgl / ck / diy / fa / ic / jp / lit / sci / vr / vt ] [ index / top / reports ] [ become a patron ] [ status ]
2023-11: Warosu is now out of extended maintenance.

/vt/ - Virtual Youtubers

Search:


View post   

>> No.67215053 [View]
File: 419 KB, 1678x1620, sora_stats.jpg [View same] [iqdb] [saucenao] [google]
67215053

Hello /#/ bros, I tried to play around with VStats to motivate myself in learning data analysis. I'm pretty much new to this sort of thing so just call me out if I made a mistake in what I did.

I attempted to use an AR(1) model for CCV graphs that look "nice" enough. Cooked some code for this and ran it through some streams and the results were pretty decent.

For example, let's consider Daisenpai Sora's recent karaoke stream ( https://www.youtube.com/watch?v=ua8XotCYB80 ). Under the AR(1) model I can say that
>on average, Sora was able to maintain the attention of 90% of her audience after every minute
>on average, Sora was able to attract around 611 curious viewers per minute to her karaoke stream

The relevant graphs are in picrel. The only important ones are the top two, the bottom ones are just for studying the residuals in the AR(1) model on the top right.

Of course, I am not saying that this is totally correct, but at least it gives us a more numerical idea of how a stream performs. I could also use the program to predict max. CCV, which I plan to do tomorrow.

Navigation
View posts[+24][+48][+96]