Out of curiosity, what is stopping a rioter from just making this into a skin? Or a very closely related one. I feel like we see so many good skin ideas like this but rarely ever see them picked up by riot nowadays.
I'm not in Legal or an artist, but iirc there are legal implications about using a skin from a non-Rioter.
If you experience any other issues, gimme your frustrations in this thread and we'll work to help you out. Preferably watch mission issues, but if you want help with literally anything else, sure why not I'll give it a shot.
I have a sideways related question. On the Rewards page it indicates I have watched 458 games for 434 hours watched. Does the latter figure only include game time, or does it include the desk segments/PGL, etc?
The latter figure is represented by the former in your question.
(only game time, no show segments)
This
When logged in to the website, once i go to the VOD section the website will automatically log me out and redirect me to watch it logged out. Is this a known issue or a way to correct this?
I have not heard of that bug, it sounds unpleasant :(
Have you tried incognito mode / a different browser to see if it is an issue with your specific config?
Maybe because it’s something more culturally supported/accepted, which encourages more people to get into it? Kind of like how 44% of NHL players are Canadian.
We made something similar to this at the hackathon 3 years ago. We found champions had minimal impact, compared to other variables (mostly gold + objectives). Does this match with your model?
Link for anyone interested: https://benfradet.github.io/blog/2016/10/22/riot-games-2016-hackathon
Hey that’s a pretty awesome write up. I’ve been dogfooding our developer API to try and make some pre-built graph datasets to show player associations, but scraping matches upon matches is a bit slow.
I mean sure, but some champions don’t have up to date recommended pages, and for new players this can be really annoying. E.g. a lot of on hit champions still have like crit in their page when they build all on hit
AFAIK this is the main reason it’s been talked about. Recommended weapons are still statically set during each build of the client. The more we can automate, the better.
I’m sure there will be a glitch though where it recommends 6 tears if you’re playing Draven and die to first blood.
Linear mixed models, 412 variables, delta, 0.
You getting hard yet?
Just don’t over-fit the model.
is there any chance at all that this software could be used to actually punish trolls? for example, if someone plays ap poppy mid with ghost cleanse and never last hits, under the current system, a report would do nothing. would this software recognize that these choices created such a detriment to that team that, coupled with a report, this player would be punished?
Maybe, but I think it would be pretty hard. We always want players to be free to explore new things, and that you don’t just have to play the current meta. A tanky midlaner not last hurting? Sounds like it could be an attempt at a new funnel strat. Hiesendong taking smite for the health regen off raptor camp was pretty creative before it got nerfed.
You (or rather the data science team) should read about Sugihara's work
Can’t speak about the DS team, but it probably doesn’t help me directly. My team has to transmit 10s of gigs of data across the globe per second, and then sort/store/catalog it all in real time.
Might read it for fun though!
You had me at MapReduce
Sure MR is fun, but when you start getting into stream processing? That’s when my ears really perk up.
This is a super common problem in social science research. It's actually very solvable in this application, I've actually thought a lot about how it could be done. Hire me and I'll tell you ;)
That makes you part of the problem then
nigga what the f*ck is your issue?
He never said it was ranked and it seems he didn't screenshot either
It was ranked, and in no way do i wanna waste at least 15 minutes of my life and my precious lp just so i could report some annoying f*ckers
Way to just sneak in nonchalant like
That’s me! Mr. Sneaky Rito Man!
I just get excited anytime data gets talked about.
How accurate it is?
Honestly can’t speak to it. Stuff like this at Riot is split across two larger teams, one group doing data collection and storage (my team), and data scientists who do machine learning later on. I know there have been many variations of models over the years, and the hard part is to determine causality (Ex: did you win because you secured Baron, or did you get Baron because you had such an insane lead and were going to win anyways?)
They probably trained a machine learning model using gold lead, objectives, and team comp
You are correct.
Read moreToday's PMTs
RY vs DWG / MEGA vs HKA / DWG vs FLG / MEGA vs LK / RY vs FLA / HKA vs LK
Yesterday's PMTs
Frosk is a color commentator!
Try not dodging next time
so having 2 people wanting mid in my team and no support in my team, someone banning my champ and everyone flaming in champ select is ok to you? when i have games as those i just dodge to save my nerves...
devtrackers.gg
