Alert Statistics And Analysis

[SoaringEagle]

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.

They never said the rewards are to be evenly distributed. Rather, they said it is completely random. All of the current data should be shown and taken note of.

Just guessing, but I bet we have one month of data to show now, at least. EDIT: Yep, Twitter Alerts began on the 12th of April, so more than one month of data.

O, and also please don't include the 12hr, 900m Orokin items, as they were not part of the random Alert system.

Edited May 23, 2013 by SoaringEagle

[dunvi]

They never said the rewards are to be evenly distributed. Rather, they said it is completely random. All of the current data should be shown and taken note of.

Just guessing, but I bet we have one month of data to show now, at least. EDIT: Yep, Twitter Alerts began on the 12th of April, so more than one month of data.

O, and also please don't include the 12hr, 900m Orokin items, as they were not part of the random Alert system.

 

I can filter those out, but again we're looking at 17 or 18 datapoints total. There's nothing to conclude from that.

[Meltina]

I do wish they started the twitter earlier before we just had the alert forums and i dont expect anyone to go through and manualy catalog all the ? alerts from that. But you can see places where its weeks between catalysts so that would be with a few more months of data you still only get about 13 more data points or so. If your taking the data from the part of the beta.

[dunvi]

I do wish they started the twitter earlier before we just had the alert forums and i dont expect anyone to go through and manualy catalog all the ? alerts from that. But you can see places where its weeks between catalysts so that would be with a few more months of data you still only get about 13 more data points or so. If your taking the data from the part of the beta.

 

I agree, and I considered scraping the alert forum here, but then I decided I could focus on making the future stats better :P

 

 

This was thrown together very very quickly and roughly. Potatoes (no differentiation between reactors and catalysts) by start time, minus the 2 half-day ones. I see no trends, and I also see not enough data. Exactly as I predicted.

[Meltina]

Just a big lack of potatos at prime time. With more data we can see if its just RNG hating us or if the odds are weighted to off hours of the game. Right now anyone who works 9-5 plays warframe then goes to bed to go to work the next day has been screwed in the distribution. Which as those people are with jobs i can see the possiblity of making alerts rarer in that time peroid to promote plat purchases.

[Kahruvel]

Holy crap, I was actually working on a project just like this today. You beat me to it. I was going to include a list of average BP drops per day, including peak and low (I can tell you it's zero, because a few days had no BP drops. Very few though).

 

Good show.

[dunvi]

Holy crap, I was actually working on a project just like this today. You beat me to it. I was going to include a list of average BP drops per day, including peak and low (I can tell you it's zero, because a few days had no BP drops. Very few though).

 

Good show.

 

Ooh, yes, that stat is in my list of things to do, I just haven't gotten to it yet. Any others?

[potatofamine]

From that graph, 8 potatos have been within a 4.5 hour span of time, and the other 9 potatos spread out between the other 19.5 hours of the day. Sure, if you had 6 more months of data maybe they would be more evenly spread out, but the fact remains that half of the potatos so far have been in the same 4.5 hour time period.

[dunvi]

From that graph, 8 potatos have been within a 4.5 hour span of time, and the other 9 potatos spread out between the other 19.5 hours of the day. Sure, if you had 6 more months of data maybe they would be more evenly spread out, but the fact remains that half of the potatos so far have been in the same 4.5 hour time period.

 

... yeah you've never taken a statistics class, have you.

[SoaringEagle]

I can filter those out, but again we're looking at 17 or 18 datapoints total. There's nothing to conclude from that.

I have to stress the period of time in which the data has been collected. If we went by each week, I'd say that's too short of a time span to make any conclusions. The more time we have to collect data the closer we can make a conclusion that is sound.

One month is plenty of time.

 

I agree, and I considered scraping the alert forum here, but then I decided I could focus on making the future stats better :P

 

 

This was thrown together very very quickly and roughly. Potatoes (no differentiation between reactors and catalysts) by start time, minus the 2 half-day ones. I see no trends, and I also see not enough data. Exactly as I predicted.

Now from the graph, spanning one month plus one week plus three days, we can conclude that it is possible for someone who has three hours or less to play each day may never see an Orokin blueprint.

The chance of seeing an Orokin blueprint is further reduced when, we can assume, people do not have the same daily routine all seven days of the week. We can throw in other games and activities which would not allow them to participate, in that three hour timespan, and you have a very unrewarding system.

40 days at 3 hours per day = 120 hours of play time. That's a lot of time played, and they're missing out just because of the time of day they play. That's not fair, that's not right.

[Judopunch]

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.

Ya when you assume players have a 5 hour window, out of 24 hours to get an alert with a ? reward, then a useful ? reward.... Starts to look grim :(

[Argoms]

... yeah you've never taken a statistics class, have you.

His point still stands. He's not saying that this indicates that it's skewed in any way, he's saying that the current sample shows that some people have had drastically more opportunities than others so far.

Edited May 23, 2013 by Argoms

[RoundedPlanet88]

The ? system is a RNG system as a WHOLE, i.e., only considering individual rewards is ultimately an exercise in futility. I could look at ANY of the specific rewards, and say that for somebody who has played 500 of the past 600 hours, has missed all of that specific type of reward. simply because they aren't on at the "magic times". My point being that ONLY considering potatoes, or any other sub-section of rewards, and then applying imagined trends taken from an ENTIRE SEVENTEEN DATAPOINTS, and then applying it to a larger system to be more than a bit............................you know what, nvm.

You guys are TOTALLY RIGHT. Seventeen data points is MORE THAN SUFFIECENT TO DRAW TOTALLY LEGIT CONCLUSIONS FROM, Especially from a sub-section of the actual whole dataset.

[Naed]

The ? system is a RNG system as a WHOLE, i.e., only considering individual rewards is ultimately an exercise in futility. I could look at ANY of the specific rewards, and say that for somebody who has played 500 of the past 600 hours, has missed all of that specific type of reward. simply because they aren't on at the "magic times". My point being that ONLY considering potatoes, or any other sub-section of rewards, and then applying imagined trends taken from an ENTIRE SEVENTEEN DATAPOINTS, and then applying it to a larger system to be more than a bit............................you know what, nvm.

You guys are TOTALLY RIGHT. Seventeen data points is MORE THAN SUFFIECENT TO DRAW TOTALLY LEGIT CONCLUSIONS FROM, Especially from a sub-section of the actual whole dataset.

Sorry mate, but logic doesnt work here, at least on this subject :/