EPTeSports Interactives: IEM Katowice 2015 - Page 2
| Forum Index > General Forum |
|
amazingxkcd
GRAND OLD AMERICA16375 Posts
| ||
|
RhYS331
7 Posts
 | ||
|
Gwavajuice
France1810 Posts
On May 14 2015 21:36 malahay wrote: Hi, I'm Efe the author of the article. The visualization started with three questions in mind: 1) Do viewerships increase depending on the teams that are played? 2) How many viewers do you lose when a break kicks in, or the analysis desk takes over? 3) How much time was spent on games vs. other stuff (analysis & breaks)? Hopefully, we manage to give an answer on these problems. Yurie in the comment above had a very interesting observation: So hopefully, you'll be able to discover interesting insights as well! 1) yeah but it you only give "much more viewers" as a result. did you effectively a model with teams as factors and viewers as outcome, did you analyse each teams impact (like in a tukey test for example) what is the IC you obtained, did you run a fisher test or another method? did the model take time of the day as a factor? 2) did you try to estimate a percentage of loss? what did you obtain? with what IC? was it dependant on team/time/stage of competition? 3) Ok with this one, not totally sure with what to do with this information though Edit : I'm not saying you're work is not good, just curious to see what happens when you go beyond the raw data... | ||
|
Circumstance
United States11403 Posts
| ||
|
ProBell
Thailand145 Posts
If you want pure viewership and fun enjoyable environment you want casters/players like MC/Dragon/Parting. And the "serious" tasteless/artosis @ non-GSL events is not nearly as fun/funny as the GSL tastetosis. Just my opinion as a long time sc2 viewer anyway. | ||
|
malahay
5 Posts
On May 15 2015 10:34 Circumstance wrote: Is Mr. Karakus involved with the upcoming ESL movie in any way? Just wondering. haha no I am not  I am about to graduate from grad school. I just built this interactive because I thought it could be a fun project. | ||
|
nachtkap
Germany197 Posts
On May 14 2015 16:13 Zealously wrote: Now, as part of his work on a masters' thesis, he set out to analyze how viewer counts fluctuate during a broadcast, and to determine the factors that influence these fluctuations. Clicking the link below will take you to a dedicated page for his project. Wait, what? You actually managed to convince, what I assume must be, Computer Science Professor to let you do master thesis about esports. I certainly hope you never try sell me your car  .Joking aside I'm seriously interested. I assume you had to write a proposal for this. Was all that really mattered that you had a way to gather enough (viewership) data, a difference in content (games, breaks/ads, analysis) and the questions to ask about it? Or did you have to obfuscate what you would be analyzing with something like 'effects of breaks and ads on streamed content'? | ||
|
Gwavajuice
France1810 Posts
On May 15 2015 17:46 nachtkap wrote: well this is awesome... an actual analysis... with actual data... about esports =) Wait, what? You actually managed to convince, what I assume must be, Computer Science Professor to let you do master thesis about esports. I certainly hope you never try sell me your car .Joking aside I'm seriously interested. I assume you had to write a proposal for this. Was all that really mattered that you had a way to gather enough (viewership) data, a difference in content (games, breaks/ads, analysis) and the questions to ask about it? Or did you have to obfuscate what you would be analyzing with something like 'effects of breaks and ads on streamed content'? Well in my days, you could take any subject that had data qualified enough to let you use enough analysis and visualition tecniques. It didn't matter much if the subjects were customers, cars, crops, hair styles or bacteriae | ||
|
Japetto
Australia1 Post
I was wondering how accurate and reliable things are and if this trend is seen in other events around the world or even more regular tournaments like proleague or GSL. You have stream numbers here but as with all online content it is hard to show how many people are 'truly' watching. I know I'm guilty of wandering off/away from the PC while breaks are on, coming back for the last of the analysis or even the beginning of the game as well factoring multiple viewers (ie barcrafts/friends). I am unsure as to how commercial television obtains it's numbers perhaps there would be a similar equation that could be applied here. Another interesting little number would be being able to see where viewers are located across multiple regions and multiple events. Oh the possibilities are endless. Thanks for the info. The twitch revenue was particularly eye opening. | ||
|
malahay
5 Posts
On May 15 2015 17:46 nachtkap wrote: well this is awesome... an actual analysis... with actual data... about esports =) Wait, what? You actually managed to convince, what I assume must be, Computer Science Professor to let you do master thesis about esports. I certainly hope you never try sell me your car .Joking aside I'm seriously interested. I assume you had to write a proposal for this. Was all that really mattered that you had a way to gather enough (viewership) data, a difference in content (games, breaks/ads, analysis) and the questions to ask about it? Or did you have to obfuscate what you would be analyzing with something like 'effects of breaks and ads on streamed content'? Hi Haha I think there was a misunderstanding, I did my thesis on certain aspects of esports but this project was just purely for fun. So no proposal or IRB approval was necessary for the project. The data collected didn't need any sort of obfuscation because it's public. You can make requests to the Twitch API (https://github.com/justintv/Twitch-API) to get viewership numbers. Hope this helps! On May 16 2015 09:58 Japetto wrote: This is some good stuff even if it isn't directly looking at SC2. I was wondering how accurate and reliable things are and if this trend is seen in other events around the world or even more regular tournaments like proleague or GSL. You have stream numbers here but as with all online content it is hard to show how many people are 'truly' watching. I know I'm guilty of wandering off/away from the PC while breaks are on, coming back for the last of the analysis or even the beginning of the game as well factoring multiple viewers (ie barcrafts/friends). I am unsure as to how commercial television obtains it's numbers perhaps there would be a similar equation that could be applied here. Another interesting little number would be being able to see where viewers are located across multiple regions and multiple events. Oh the possibilities are endless. Thanks for the info. The twitch revenue was particularly eye opening. I think all these questions are right on point. Unfortunately, there is no real way of measuring if someone is "really" watching  or just have the stream on in a tab or just listening. Perhaps if you work at Twitch you might be able to see if the stream is right now on full screen and judge that the person is really watching that way.The way TVs measure things are even more inaccurate as far as I know (http://en.wikipedia.org/wiki/Nielsen_ratings) The location would again be definitely fun to see, but that data is private and not provided by the API for now Thanks for the great thoughts everyone! | ||
|
nachtkap
Germany197 Posts
On May 16 2015 13:13 malahay wrote: Hi Haha I think there was a misunderstanding, I did my thesis on certain aspects of esports but this project was just purely for fun. So no proposal or IRB approval was necessary for the project. The data collected didn't need any sort of obfuscation because it's public. You can make requests to the Twitch API (https://github.com/justintv/Twitch-API) to get viewership numbers. Hope this helps! I did mean the thesis (proposal). It was a tongue in cheek question. I wouldnt have thought that a thesis proposal covering esport would get approved. What is the thesis about and is available anywhere (arxiv, pubmed, or something similar)? | ||
|
malahay
5 Posts
On May 17 2015 23:04 nachtkap wrote: I did mean the thesis (proposal). It was a tongue in cheek question. I wouldnt have thought that a thesis proposal covering esport would get approved. What is the thesis about and is available anywhere (arxiv, pubmed, or something similar)? It's not out yet unfortunately. I think the university makes it public two years after submission I was looking into comparing certain aspects of esports and traditional sports. The two big visualizations that were part of my thesis were:* LCS Fandom Shift: http://efekarakus.github.io/esports/fandom/ (for LoL) * Twitch revenue: http://efekarakus.github.io/twitch-analytics/#/revenue And some other things like what retired professionals do that are just static images and not interactive on the web. | ||
|
opisska
Poland8852 Posts
| ||
| ||
