Commander
Join Date: Oct 2012
Posts: 289
# 131
02-23-2013, 12:30 AM
There are no Purple XII booster modulators, I have never crafted one ever. Or ever seen one on exchange
Career Officer
Join Date: Jun 2012
Posts: 164
# 132
02-23-2013, 06:56 AM
So, I did the obvious thing and added the totals of the two large samples together. There are a total of 768 consoles in them. In a couple cases, the subtotals given by AdmGreer did not agree with his itemized list, so I assumed the itemized list was right.

One odd thing is that on page 1 of this thread, Bort said there are 16 Eng, 10 Sci, and 16 Tactical Consoles you can make.

In fact, that does not agree with the number of actual console types seen.

There are 15 Eng consoles, if you count Booster Modulator, or 14 if you don't. If there's a 16th Eng console, it didn't appear in the sample of 271 Eng modules crafted.

There are 12 Sci consoles, not 10. According to AdmGreer's numbers, he did not get a Biofunction Monitor or a Shield Emitter Amplifier.

There are, in fact, 16 Tac consoles; AdmGreer did not get a Phaser Relay, nor did he list the Zero Point Quantum Chamber.

14+12+16 does add up to the advertised number of 42 possible outcomes.

If there are consoles in the loot table (e.g. Booster Modulator Mk XII, or the mysterious 16th Eng console) that don't exist, that might be a pretty big deal. References in code, or data tables, to things that don't exist tend to lead to unpredictable results...

If we assume Booster Modulators aren't in the table, the chi^2 for the combined sample being uniformly distributed among the 42 console types that appear in at least one sample is 137 for 41 DOF. If the consoles are truly uniformly distributed, the odds against observing an outcome this uneven (or worse) are about 400 billion to 1.

For the combined data, the distribution among Eng/Sci/Tac consoles is also very far from expected:

Eng: 271 observed, 256.0 expected
Sci: 272 observed, 219.4 expected
Tac: 225 observed, 292.6 expected

This corresponds to a chi^2 of 29 for 2 degrees of freedom; the chance of observing such an outcome due to chance if the consoles are distributed 14:12:16 is also extremely small: about 1-in-2 million.

There is a possible selection bias at work here. AdmGreer posted because he observed an extremely non-uniform distribution. So it is interesting to look at only DeusEmperor's data.

For the Eng/Sci/Tac consoles, Deus alone has chi^2 6.9 for 2 degrees of freedom. If the consoles are distributed 14:12:16, the chance of that is about 3%. Small, but not that small.

For the frequency of individual consoles, the chi^2 for Deus alone is 62.6 for 41 degrees of freedom. The chance of this outcome from a uniform distribution is slightly less than 2%. About half of the chi^2 is due to one console (Ablative Hull Armor). Except for that one outlier, Deus' data would be reasonably consistent with uniformity. But the chance of such an outlier if the consoles are uniformly distributed is pretty small (as reflected in the chi^2 probability).

It doesn't look to me like the distribution (which was presumably sampled over a fairly long period of time), was uniform at all times.

That doesn't mean that it is necessarily non-uniform for consoles crafted *now*. Just speculating, but a problem lasting for only a few weeks a year or so ago, that led to a few types of consoles being heavily over-represented during that time, would be enough to account for the non-uniformity of the cumulative data. Something weird that resulted in 10 extra copies of Ablative Armor over a relatively short time would be enough to explain Deus' data, for instance.

I wonder if a more controlled sample could be tested by repeated character copies to Tribble...

Last edited by lagunad; 03-03-2013 at 09:42 PM.
Lt. Commander
Join Date: Jul 2012
Posts: 138
# 133
02-23-2013, 08:15 AM
Anyone who's ever done large quanties of these has known this is rigged; if you believe otherwise, you are a special type of naive... Its in the interest of business that premium consoles like purple Mk XII Field Generators and damage consoles stay very, very rare, that way if you want one your best bet is to just BUY KEYS and sell them for the amount of EC you need...

I look forward to seeing those test results derped away...
Rihannsu
Join Date: Jun 2012
Posts: 12,514
# 134
02-23-2013, 10:31 AM
I collated the Data from the two sets together and noticed that the two sets of data have a lot of variation. This lends credence to the idea that it's selection bias at work. the number on the left is the one from the OP, the second is the one from last page:

ENG:
Ablative Hull Armor (5)22
Booster Modulator (0)0
Diburnium Hull Plating (8)11
Electroceramic Hull Plating (5)8
Emergency Force Fields (9)2
EPS Flow Reg (4)10
Field Emitter (19)5
Injector Assembly (20)12
Monotanium Alloy(4)13
Neutronium Alloy (2)15
Parametallic Hull Plating (8)11
Plasma Distribution Manifold (16)16
RCS Accelerator (1)12
SIF Gen (9)10
Tetraburnium Hull Plating (10)4

SCI:
Biofunction Monitor (0)12
Countermeasure System (19)5
Emitter Array (8)9
Field Generator (1)9
Flow Capacitor (13)14
Graviton Generator (11)13
Inertial Dampeners (14)9
Particle Generator (8)15
Power Insulator (20)12
Sensor Probes (25)7
Shield emitter amplifier (0)11
Stealth Module (27)10

TAC:
Ambiplasma Envelope(4)8
Antiproton Mag Regulator (2)4
Chroniton Flux Regulator (4)8
Directed Energy Distribution Manifold (18)5
Disruptor Induction Coil (2)6
Phaser Relay (0)4
Photon Detonation Assembly (6)9
Plasma Infuser (4)7
Polaron Phase Modulator (6)10
Prefire Chamber (11)10
TCD Subspace Infuser (3)10
Tetryon Pulse Generator (4)14
Transphasic Compressor (7)9
Variable Geometry Detonators (8)7
Warhead Yield Chamber (17)8
Zero Point Quantum Chamber (0)10

this time the spread is 0 - 37, and now to graph:
00-01 -
02-03
04-05 -
06-07 -
08-09 -
10-11 -----
12-13 ------
14-15 ----
16-17 -----
18-19 ----
20-21 -
22-23 ---
24-25 ---
26-27 --
28-29
30-31
32-33 ----
34-35 -
36-37 -

The expected average here is 18. It's kinda odd seeing so many numbers in the 30s. but, as was said earlier, selection bias.

But for this to be truely meaningful, we need a larger sample size. 42^2 would be enough for a meaningful sample.
HAIL HYDRA!

-=-=-=-=-=-=-=-=-=-
I can haz joystick!
MMOs aren't charities. Corporations are supposed to make a profit. It's what they do.

Last edited by markhawkman; 02-23-2013 at 10:35 AM.
Lt. Commander
Join Date: Jun 2012
Posts: 135
# 135
02-23-2013, 10:45 AM
Quote:
It doesn't look to me like the distribution (which was presumably sampled over a fairly long period of time), was uniform at all times.
Well I started keeping track when I read this thread and created a spreadsheet on:

2/6/2013 7:21pm MST

And I crafted 406 consoles until 2/22/2013 I'm still keeping track. So this data from me is 16 days worth of crafting so less than a month.

Quote:
14+12+16 does add up to the advertised number of 42 possible outcomes.
Yeah guess we need bort or someone to add that booster modulator and I do wonder what the mysterious 16th eng console is. lol.

Maybe an update or something removed the booster modulator thing somehow.


Edit: Also the reason why my distribution is more uniform could be because I figured out Critical Traits do matter. I have 3 toons with efficient, and 4 toons with Cunning. I've also found out that non engineers do get a 2.199% higher chance on average of getting a purple. Sometimes it jumps high and somtimes jumps lower than that. I'm also considered lucky by my friends with this mission while they don't get good stuff I do.

Maybe because I'm "lucky" I'm not a good data source, and also because I've been running 4 toons that have 5 doffs each with cunning and 3 toons with 5 doffs with efficient.
http://sto-forum.perfectworld.com/sh...d.php?t=490421 We want this help build support!

Last edited by deusemperor; 02-23-2013 at 11:04 AM.
Career Officer
Join Date: Jun 2012
Posts: 164
# 136
02-23-2013, 04:54 PM
Quote:
Originally Posted by markhawkman View Post
But for this to be truely meaningful, we need a larger sample size.
We have a 7 standard-deviation effect for the combined sample. If you don't think that is "truly meaningful", perhaps you should retake Stats 101...

Quote:
Originally Posted by markhawkman View Post
42^2 would be enough for a meaningful sample.
There is nothing special about a sample size of 42^2.
Rihannsu
Join Date: Feb 2013
Posts: 276
# 137
02-23-2013, 05:00 PM
Awesome stuff.... Great job. I wish the site had an official section for intel such as this for the community. I also think people that such tests should get a special title in game if the data submitted is valid as a special recognition of their service to the community.
Rihannsu
Join Date: Jun 2012
Posts: 12,514
# 138
02-23-2013, 08:11 PM
Quote:
Originally Posted by lagunad View Post
We have a 7 standard-deviation effect for the combined sample. If you don't think that is "truly meaningful", perhaps you should retake Stats 101...
7 standard-deviation effect? no clue what that means...
Quote:
There is nothing special about a sample size of 42^2.
Actually, there is.... Assuming there are in fact 42 possible outcomes, 42^2(or 1764) is the number of times you need to do it in order to have an average that is equal to the number of possible outcomes. It's also the point where the sample size becomes large enough for sampling "error" to become less important.
HAIL HYDRA!

-=-=-=-=-=-=-=-=-=-
I can haz joystick!
MMOs aren't charities. Corporations are supposed to make a profit. It's what they do.
Lt. Commander
Join Date: Jun 2012
Posts: 135
# 139
02-23-2013, 09:18 PM
Well I guess I'll just have to make another toon and get up and crafting that way I can do 24 consoles a day to get up to 42^2 number it probably take a bit.

1764-426=1338 more consoles to go. At 24 consoles a day it will take 55.75 days. so a little over 2 months from now for me to get your desired sample size.

My data is correct I have the spreadsheet opened and add to it 2-4 times a day 7 consoles at a time. I see what I got in game and put a number on the console on the spreadsheet.

Maybe my data is more uniform as I on average make 21 consoles a day right now, while the original post did it off and on as he eventual got them from his fleet mates.
http://sto-forum.perfectworld.com/sh...d.php?t=490421 We want this help build support!
Career Officer
Join Date: Jun 2012
Posts: 164
# 140
02-24-2013, 12:28 PM
Quote:
Originally Posted by markhawkman View Post
7 standard-deviation effect? no clue what that means...
So you have "no clue" about one of the most elementary concepts in statistics? OK.

Quote:
Originally Posted by markhawkman View Post
Actually, there is.... Assuming there are in fact 42 possible outcomes, 42^2(or 1764) is the number of times you need to do it in order to have an average that is equal to the number of possible outcomes. It's also the point where the sample size becomes large enough for sampling "error" to become less important.
Actually, there isn't. See above.

According to your "rule", since there are 2 possible outcomes to a coin flip, 2^2 = 4 coin flips is a large enough sample to tell whether the coin is "fair" (has equal probability to come up heads or tails).

But in four flips, a fair coin will give four heads OR four tails 12.5% of the time. So 2^2 = 4 flips is in no way a sufficient sample size and clearly your "rule" is nonsense. Whoever told you that "rule" was either simply making it up, or you misunderstood them.

(I have a PhD in experimental particle physics and teach a statistics course at a major university, by the way).
Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is Off
HTML code is Off


All times are GMT -7. The time now is 11:10 PM.