RandomSeed test (was:randomly order a list)

[dunbarx at aol.com]

Dar.


Yes, the randomseed is misused here. Parameter is better.



I assumed that it made no difference if I sorted the original text, or with a sorted version of that text. Random is random, right? An example of shoot-from-the-hip posting to the list.


It seems that it does, and I am not sure why. With a low sort parameter, there is a distinct favoring of the original text, and a distinct disfavoring of the reverse of that text, The other sorted results are all about where they should be, 16% of the values.



I am sure there is an explanation for all this. There is definitely something to what Chris said. The original text occurs far more often than it should. I think there was discussion about this a while back, and I have forgotten what the upshot was.


Bottom line, though, use a big number.


Craig




-----Original Message-----
From: Dar Scott <dsc at swcp.com>
To: How to use LiveCode <use-livecode at lists.runrev.com>
Sent: Wed, May 22, 2013 5:26 pm
Subject: Re: RandomSeed test (was:randomly order a list)


There is a potential confusion in the term randomSeed.  We might be using it two 
ways.

There is the randomSeed property, the state of the random number generator.  

And there is the argument to the random function.


And...
Shouldn't the 'get fld 1' be inside the loop?  (Or initialized the same way each 
time.) You are using the previous sorted 'it' each time.  Perhaps that is what 
is doing the scrambling.

This is my test:

on mouseUp
   set the cursor to watch
   put 0 into a1
   put 0 into a2
   put 0 into a3
   put "a" & lf & "b" & lf & "c" into x
   repeat 1000000 times
      put x into myVar
      --sort lines of myVar by random(the number of lines of myVar)
      sort lines of myVar by random(999999)
      if line 1 of myVar is "a" then add 1 to a1
      if line 2 of myVar is "a" then add 1 to a2
      if line 3 of myVar is "a" then add 1 to a3
   end repeat
   put a1 && a2 && a3 into field 1
end mouseUp

I got different results depending on which line is commented out.

Dar


On May 22, 2013, at 2:46 PM, dunbarx at aol.com wrote:

> Sorting by a randomSeed greater than the number of elements of a sample does 
not help.
> 
> 
> With the following in a fld 1:   aaa,bbb,ccc
> 
> 
> and this in a button:
> 
> on mouseUp
>   get fld 1
>   repeat  1000
>      sort items of it by random(3)
>      put it & return after lowSeed
>   end repeat
> 
>   repeat  1000
>      sort items of it  by random(3333)
>      put it & return after highSeed
>   end repeat
> 
>   repeat with y = 1 to the number of lines of lowSeed
>      add 1 to lowSeedCounter[line y of lowSeed]
>   end repeat
> 
>   repeat with y = 1 to the number of lines of highSeed
>      add 1 to highSeedCounter[line y of highSeed]
>   end repeat
> 
> end mouseUp
> 
> 
> The distribution of the two groups is about the same. 
> 
> 
> Craig Newman
> 
> 
> 
> 
> -----Original Message-----
> From: Dar Scott <dsc at swcp.com>
> To: How to use LiveCode <use-livecode at lists.runrev.com>
> Sent: Wed, May 22, 2013 3:59 pm
> Subject: Re: randomly order a list
> 
> 
> Try replacing 'the number of lines of myVar' with some huge number you think 
is 
> allowed for random(), say 999999.
> 
> With small numbers you are likely to get the same number for different lines 
and 
> perhaps sort does not change order in that case. 
> 
> Dar
> 
> 
> On May 22, 2013, at 11:59 AM, Chris Sheffield wrote:
...
>> 
>> 	sort lines of myVar by random(the number of lines of myVar)
>> 
...
>> 
>> --
>> Chris Sheffield
>> Read Naturally, Inc.
>> www.readnaturally.com
> 

_______________________________________________
use-livecode mailing list
use-livecode at lists.runrev.com
Please visit this url to subscribe, unsubscribe and manage your subscription 
preferences:
http://lists.runrev.com/mailman/listinfo/use-livecode