dunvi Posted May 22, 2013 Share Posted May 22, 2013 (edited) Okay, I've only just started this, but I'm working on some simple stuff to just grab all of the alert tweets and calculate some stats on them. The code is on my github page, and what I've got so far can be found in the file stats.txt, and in the folder images. I'm working on getting a live server running this so I can pull in updates hourly - I'm having mongodb problems, so right now it only updates whenever I run the update script on my local machine. Once that's going I can make a simple webpage to show the graphs automatically, but it's not my main concern right now. Some statistics people keep on asking for, as of May 22, 2013 at 7:00pm pacific time: 1921 alerts on twitter ~14% of all alerts have ? awards ~9.4% of all alerts have an artifact award ~4.5% of all alerts have a blueprint award (minus potatoes) ~0.9% of all alerts have a potato award (either type) There have been 18 potatoes since the twitter feed was started There are not enough datapoints to do any real analysis on more specific ? alert rewards. Finally, start time distribution: That would be the distribution of warframe alerts by start time, divided into 30 minute intervals (GMT times, I will care about fixing the labels later :p). If you're curious about what it means, it's proof that the RNG is actually quite RNG (though that said, the 15-minute interval version of this image is interesting...) so honestly, complaining about alerts being in the middle of the night doesn't really hold a lot of water for me. More stats can be found if you follow the links, and I will add more as I think of them/procrastinate. Feel free to throw me ideas for other things to look at or fix - I'm not a statician, and I'm not entirely sure how I passed prob-stats :p. If you want to see my todo list, it's here. Edited May 23, 2013 by dunvi Link to comment Share on other sites More sharing options...
Tabtree Posted May 22, 2013 Share Posted May 22, 2013 can you calculate the chance that an alert will have a ?, based on past alerts that had a ? compared to normal alerts??? Link to comment Share on other sites More sharing options...
KvotheTheArcane1 Posted May 22, 2013 Share Posted May 22, 2013 Just a small complaint, can you get labels on sides so we can tell what means what? (I assume left is num of alerts, bottom is time?) Link to comment Share on other sites More sharing options...
MaduSciuntistu Posted May 22, 2013 Share Posted May 22, 2013 and all the alerts are for 2k Link to comment Share on other sites More sharing options...
Archistopheles Posted May 22, 2013 Share Posted May 22, 2013 Even though I do not care about Alerts being "good" or "bad" for the game, you cannot base your opinion on the fact that everyone is treated the same. That logic would dictate that as long as I hate everyone the same, it is OK that I hate everyone because it's equal hate. Link to comment Share on other sites More sharing options...
WhiteCr0w Posted May 22, 2013 Share Posted May 22, 2013 Nice job. This should prove interesting over the course of time, but RNG is RNG. Link to comment Share on other sites More sharing options...
dunvi Posted May 22, 2013 Author Share Posted May 22, 2013 can you calculate the chance that an alert will have a ?, based on past alerts that had a ? compared to normal alerts??? in stats.txt: artifacts/blueprints/potatoes frequencies: 0.0936682365254 0.0486656200942 0.0094191522763 less than 15% of all missions are ? alerts. Just a small complaint, can you get labels on sides so we can tell what means what? (I assume left is num of alerts, bottom is time?) :p I'm learning matplotlib new (I do have some experience with matlab so it should be easy... but I also despise matlab >.<) so it's on my todo list. It was hard enough to get the x axis on that picture to show tiems at all. and all the alerts are for 2k YUP. Link to comment Share on other sites More sharing options...
KvotheTheArcane1 Posted May 22, 2013 Share Posted May 22, 2013 :p I'm learning matplotlib new (I do have some experience with matlab so it should be easy... but I also despise matlab >.<) so it's on my todo list. It was hard enough to get the x axis on that picture to show tiems at all. ok cool thanks Link to comment Share on other sites More sharing options...
Argoms Posted May 22, 2013 Share Posted May 22, 2013 Would it be possible to do a distribution on when a blueprint alert pops up by time? That's what I'm really interested in ^^ Also, how come potatoes are both blueprints and separate items? Did alerts once give fully built potatoes? Link to comment Share on other sites More sharing options...
dunvi Posted May 22, 2013 Author Share Posted May 22, 2013 (edited) Would it be possible to do a distribution on when a blueprint alert pops up by time? That's what I'm really interested in ^^ Also, how come potatoes are both blueprints and separate items? Did alerts once give fully built potatoes? Working on that. Potatoes show up as blueprints? ETA: You're right, I didn't notice that. May 3rd and earlier, the potatoes were tagged blueprints. I'll put in a check for that and redo the numbers. ETA2: no significant change, though given how small a percentage they were anyway I expected as much. Edited May 22, 2013 by dunvi Link to comment Share on other sites More sharing options...
Naed Posted May 22, 2013 Share Posted May 22, 2013 Even though I do not care about Alerts being "good" or "bad" for the game, you cannot base your opinion on the fact that everyone is treated the same. That logic would dictate that as long as I hate everyone the same, it is OK that I hate everyone because it's equal hate. it actually is fine, because its your choice. :) and as long as you keep that emotion to yourself, it wouldnt matter for anyone else either. So, please, throw us a different example ? Link to comment Share on other sites More sharing options...
Hyginos Posted May 22, 2013 Share Posted May 22, 2013 can you calculate the chance that an alert will have a ?, based on past alerts that had a ? compared to normal alerts??? I did that a while back. I believe it was something like 14% chance of a ?. Link to comment Share on other sites More sharing options...
Meltina Posted May 22, 2013 Share Posted May 22, 2013 Lol at the graph showing the credit award distrubution. Can you give us a graph of ? alerts in which times. Then do one with glaives. Then one with potatos. Then one with helments. Thats where players are complaining the alerts are in the morning/middle of the night. Link to comment Share on other sites More sharing options...
dunvi Posted May 23, 2013 Author Share Posted May 23, 2013 Lol at the graph showing the credit award distrubution. Can you give us a graph of ? alerts in which times. Then do one with glaives. Then one with potatos. Then one with helments. Thats where players are complaining the alerts are in the morning/middle of the night. i'm going to make a version of the above graph which colors the bars based on how much is a special award, broken down to blueprints and artifacts, but any more precision than that won't mean a thing until we have more data points. Link to comment Share on other sites More sharing options...
SoaringEagle Posted May 23, 2013 Share Posted May 23, 2013 so honestly, complaining about alerts being in the middle of the night doesn't really hold a lot of water for me. First, good job on putting this all together, and I look forward to more stats regarding the "? Alerts" and not just all Alerts in general. However, regarding your statement here, I believe another answered it already but I just want to reinforce his answer, all of the talk about Alerts is about the exclusive items that are rewarded through it, not the credits. Much of the feedback has been to improve the Alerts so they do not impose a "time of day" requirement, and we've proven that RNG can still be maintained in these scenarios. Link to comment Share on other sites More sharing options...
potatofamine Posted May 23, 2013 Share Posted May 23, 2013 What the F*** are the numbers on the left? "That would be the distribution of warframe alerts by start time" So e.g. at 9:36 there were 40 alerts? What? WHAT IS THIS A GRAPH OF? Link to comment Share on other sites More sharing options...
Argoms Posted May 23, 2013 Share Posted May 23, 2013 it actually is fine, because its your choice. :) and as long as you keep that emotion to yourself, it wouldnt matter for anyone else either. So, please, throw us a different example ? I think he meant treating everyone equally, as in "I don't care whether you murdered my parents or just saved me from a car crash, I'm going to treat you the same way." Link to comment Share on other sites More sharing options...
dunvi Posted May 23, 2013 Author Share Posted May 23, 2013 What the F*** are the numbers on the left? "That would be the distribution of warframe alerts by start time" So e.g. at 9:36 there were 40 alerts? What? WHAT IS THIS A GRAPH OF? it's a histogram of the start times. the bars are buckets of 30 minutes, so if an alert started within that 30 minute block it gets counted in that bucket. the right hand side is the number of times that have fallen inside that bucket. The important thing is that it's almost perfectly equal across all buckets. Alerts are happening equally at all times of the day, to 30 minutes of precision. Link to comment Share on other sites More sharing options...
potatofamine Posted May 23, 2013 Share Posted May 23, 2013 Ah I get it now. Pretty complex graph. I don't think there was much concern though that alerts weren't appearing on pretty consistent 30-minute intervals, but that the "good" alerts are generally appearing at inconvenient times. Link to comment Share on other sites More sharing options...
dunvi Posted May 23, 2013 Author Share Posted May 23, 2013 (edited) Ah I get it now. Pretty complex graph. I don't think there was much concern though that alerts weren't appearing on pretty consistent 30-minute intervals, but that the "good" alerts are generally appearing at inconvenient times. I've seen people complaining that all of the awards are when people are asleep. Also I haven't made the graph yet (a little busy to fight with matplotlib) but I am going to plot the ? alerts specifically by time to show that I suspect it'll evenly distributed as well. Edited May 23, 2013 by dunvi Link to comment Share on other sites More sharing options...
Argoms Posted May 23, 2013 Share Posted May 23, 2013 I've seen people complaining that all of the awards are when people are asleep. Also I have made the graph yet (a little busy to fight with matplotlib) but I am going to plot the ? alerts specifically by time to show that I suspect it'll evenly distributed as well. I'm actually going to predict that while the ?s will be evenly distributed, the actual items given will not. For example, I believe I've only missed two potatoes since I started playing because of time zones, and I've gotten 8. Link to comment Share on other sites More sharing options...
potatofamine Posted May 23, 2013 Share Posted May 23, 2013 (edited) I've seen people complaining that all of the awards are when people are asleep. Also I haven't made the graph yet (a little busy to fight with matplotlib) but I am going to plot the ? alerts specifically by time to show that I suspect it'll evenly distributed as well. When you do that I'd suggest not including artifacts, as most of the complaining is about rare blueprints, not pistol scavenger. Edited May 23, 2013 by potatofamine Link to comment Share on other sites More sharing options...
dunvi Posted May 23, 2013 Author Share Posted May 23, 2013 To both of the above, the problem is how few there have been. In my entire data set there has only been 18 potatoes total, both reactor and catalyst. I don't need to be a math major to tell you the usefulness of that (nil) bit I can get them in here :P Link to comment Share on other sites More sharing options...
Argoms Posted May 23, 2013 Share Posted May 23, 2013 I know that the number is insignificant to get any probabilities out of, but showing people that so far, the potatoes etc. have favored certain time zones with actual evidence may be useful. Link to comment Share on other sites More sharing options...
dunvi Posted May 23, 2013 Author Share Posted May 23, 2013 I know that the number is insignificant to get any probabilities out of, but showing people that so far, the potatoes etc. have favored certain time zones with actual evidence may be useful. I doubt they have. You're going to end up with less than half of my buckets colored, that's not showing favoritism, that's showing lack of data. Link to comment Share on other sites More sharing options...
Recommended Posts
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now