Re: [Bug-gnubg] gnubg.sql - stats from my database

[address@hidden]

Hallo Michael,

Thanks for your input - I too agree with Ian Shaw's 
good advice. My head tells me that the program does not 'cheat' and I 
do try to study my mistakes and become a better player almost every 
day.

But I am also intrigued by the valid question of whether I am a 
poor player with bad luck, or a poor player with normal (or good) luck. 
And the same question with respect to the CPUs dice rolls.

I find the 
'luck' calculator built into gnubg to be a little obtuse in how it 
comes to it's conclusions - a more transparent and understandable 
method would be to perhaps display the statistics I suggested.

Another 
clarifying feature would be, after I lose to gnubg (again), to be able 
to play the game again.
But we would swap the CPU's dice rolls with my 
own.
This would clearly show if gnubg would still win when you have his 
now predetermined 'lucky rolls'.

I did check out your link of 
statistics: 

>http://www.capp-sysware.com/analysis/octnov2010-dc-dicestudy.txt
it 
is excellent and very close to what I was trying to describe. Did your 
generate those statistics using your scripts and sqlite? Or were they 
extracted from another source / program?

It would be a great tool if 
something like this was built into gnubg, where you could compare your 
rolls against the CPUs for these categories - and I suspect a lot of 
players would love to be able to see their own personal stats for these 
queries.

>Unfortunately the data you seek for doing a dice study does 
not reside 
>in the database (There is no accounting of every roll, and 
the data is 
>pretty much on a general match level)

But the above 
sounds like a showstopper for what I had in mind?

I am not really a 
programmer so I doubt that I will get much further than this on my own, 
but I would like to see the scripts if you have any time to send them.


Thanks again for your help and advice.

Regards,

djskope

p.s. it is 
interesting that the expected % didn't glaringly differ from the 
observed % values in the example data.(Apart from 'Statistics for 
bringing two checkers in off the bar' against a '2-pt board' which were 
slightly high, but nothing like the voodoo dice rolls which regularly 
*seem* to take place on my laptop :)






>----Original Message----

>From: address@hidden
>Date: 15/11/2010 10:11 
>To: <bug-
address@hidden>, <address@hidden>
>Subj: Re: [Bug-gnubg] gnubg.
sql - stats from my database
>
>Howdy,
>
>I agree with what Ian Shaw 
had to say about luck and what one should 
>study to improve ones game. 
The database is not well documented but this 
>might help you a bit:
>

>The local database (gnubg.db) is a binary file so can't be queried 
directly.
>Gnubg uses SQL Lite for storing data in the database. SQL 
Lite is as the
>name suggests - a light and small footprint SQL engine. 
If you know SQL then
>things are easier.
>
>You can query gnubg.db 
outside of Gnubg itself with an SQL lite tool
>available here:
>
>http://www.sqlite.org/sqlite-3_6_18.zip>
>Documentation on its use 
here:
>
>http://www.sqlite.org/sqlite.html>
>This of course is much more 
difficult if SQL is not something you have used
>before, but I present 
it just in case.
>
>Unfortunately the data you seek for doing a dice 
study does not reside 
>in the database (There is no accounting of 
every roll, and the data is 
>pretty much on a general match level)
>

>Your request for information seems similar to the type of data that 

>might appear in a  report like this:
>http://www.capp-sysware.com/analysis/octnov2010-dc-dicestudy.txt>
>If 
this is the type of data you seek I have python scripts that work 

>under Linux (I have never tried them on windows), and they don't come 

>with any documentation, and they were crude (I tossed them together 
to 
>meet a need I had last year analyzing dice from a particular 
online BG 
>site). Effectively the scripts use Gnubg to read MAT file, 
scan through 
>every roll, categorize the roll, and produce statistical 
output. If you 
>feel adventurous I can give you the scripts, but you'd 
be on your own, 
>and if you are not a programmer they may be of little 
help to you.
>
>Michael
>
>
>On 14/11/2010 1:10 PM, address@hidden
uk wrote:
>> Hallo all,
>>
>> I've been playing GNU-BG for a few weeks 
now and am getting
>> on ok (well, winning about 2 or 3 out of 10 lol) 
but I keep feeling
>> like I would like to know more about why I lose 
so much :)
>>
>> Is it
>> possible to query the built in database?
>>

>> The stats for comparing my
>> rolls against the CPUs I would 
particularly like to see are:
>>
>> 1. Total
>> number of doubles 
rolled (per match / plus total all time) for me /
>> GNU.
>> 2. Total 
pip count per game / per match / total all time - for me
>> / GNU
>> 3. 
Total number of doubles rolled whilst on the bar (if this is
>> 
possible to create) for me / GNU
>> 4. Total number of roll attempts to

>> 'get back on' when (n=1 to 5) slots are open in opponents home area 
(if
>> this is possible to create) for me / GNU
>> 5. Anything other 
stat which
>> you think may be useful?
>>
>> 'All things being equal', 
all of the above
>> figures should more or less even up, over time.
>>

>> I've found this gnubg.
>> sql file  on my hard drive but I have no 
idea how to query it, and my
>> MS Access program will not read it.
>>

>> Any help writing out these queries
>> and briefly instructing how to 
use them would be greatly appreciated.
>>
>>
>> I would like to then be 
able to import these results into Excel to
>> create graphs etc
>>
>> I 
think that being given the transparency of these
>> figures and making 
them readily available would go a long way to
>> finally proving to 
n00bs like me that GNU-BG does not cheat, as well as
>> reassuring me 
that my computer is not possessed! :)
>>
>> Many Thanks for
>> any 
help!
>>
>> djskope
>>
>>
>>
>> 
_______________________________________________
>> Bug-gnubg mailing 
list
>> address@hidden
>> http://lists.gnu.org/mailman/listinfo/bug-gnubg>>
>
>