User Pass
Home Sign Up Contact Log In
Forum > Discuss GLB Issues With Catch22 > NOT A BUG - CLOSED > Problem with Package Auto Adjust - NOT A BUG - CLOSED
Page:
 
climberpete
offline
Link
 
Originally posted by Dr. E
Looks about right then. Try it with dice. You will get results that look similar.


Just tried it with an online random number generator. I generated a number 1 to 16 (A - P) 72 times, same as the package in question:
A - 6
B - 4
C - 4
D - 4
E - 3
F - 4
G - 3
H - 7
I - 6
J - 7
K - 1
L - 4
M - 3
N - 6
O - 7
P - 3

In this case only one data point would be considered an outlier - K, this is an expected random dataset. The package in question has 25% of it's data points (diamonds, weak overload, HB screen weak, WR screen weak) considered outliers, that points to an anomaly.
Edited by Catch22 on Nov 8, 2011 22:01:24
 
Dr. E
offline
Link
 
Originally posted by climberpete
Just tried it with an online random number generator. I generated a number 1 to 16 (A - P) 72 times, same as the package in question:
A - 6
B - 4
C - 4
D - 4
E - 3
F - 4
G - 3
H - 7
I - 6
J - 7
K - 1
L - 4
M - 3
N - 6
O - 7
P - 3

In this case only one data point would be considered an outlier - K, this is an expected random dataset. The package in question has 25% of it's data points (diamonds, weak overload, HB screen weak, WR screen weak) considered outliers, that points to an anomaly.


Wrote a little script huh. Run it a couple thousand times then. Get a good sample.

Little side note on play selection in packages. Ever consider what would happen if no play's percentage chance was met? It would have to happen eventually, causing some sort of loop. I figured Bort has something in his code that insures the last play gets selected. Something that raises the odds of a play being called the further down the choices it goes.
Edited by Dr. E on Nov 8, 2011 20:39:58
 
climberpete
offline
Link
 
Originally posted by Dr. E
Wrote a little script huh. Run it a couple thousand times then. Get a good sample.

No script, just a quick and dirty dataset.

Originally posted by Dr. E
Little side note on play selection in packages. Ever consider what would happen if no play's percentage chance was met? It would have to happen eventually, causing some sort of loop. I figured Bort has something in his code that insures the last play gets selected. Something that raises the odds of a play being called the further down the choices it goes.


If I understand you correctly you're suggesting that it selects plays like this:
Play A 15% chance, roll number 1-100, if 1-15 play fires else move on to Play B

This is an inefficient way of selecting a play because you then have to roll multiple random numbers.

A better method is to add up all the percentages unmodified Play A - 100%, Play B - 100%, Play C - 50% and roll a number 1-250. Assign blocks of numbers to each play. For example 1-100 is Play A, 101-200 is Play B, 201-250 is Play C. This is in fact how it works when selecting an output in the AI.
 
Catch22
offline
Link
 
Select plays by assigning each play a number between 1 and 100 based upon percentages (recalcs to 100 if the percentages do not add up to 100) and then picks a random number between 1 and 100 and calls the play that matches that number.

Not going to allow a test server test on this - sorry, I just don't have enough to go on to justify it. Keep running your data and get back to me when the sample size is much larger. Closing this thread now.
 
Catch22
offline
Link
 
Thread moved by moderator.
 
Page:
 


You are not logged in. Please log in if you want to post a reply.