作者: superkufu 時間: 2008-12-22 09:07 標題: 六合彩Data Mining
之前做完Data mining ......
有無人有玩過Data Mining, 覺得應用係搵六合彩有可能?
XD....
作者: alphaau 時間: 2008-12-22 09:31
search for john nash,
game theory
but you need more data then just the six/sever number!
作者: superkufu 時間: 2008-12-22 09:56
諗左ga la....
之前個final year 係做stock market 既 data mining...
想試d pattern haha
雖然無深究佢個game theory...
但係近似於probability , 甘樣可能問題.. 每一個number 既 occurence 會同其他number無關(雖然可能真係可能無關
), 但可能有相關性.. game theory 同 naive bayes 相似(?)
http://en.wikipedia.org/wiki/Naive_bayes
但係用SVM 可以solve 到個相關性 XD....
http://en.wikipedia.org/wiki/Support_vector_machine
[ 本帖最後由 superkufu 於 2008-12-22 09:56 編輯 ]
作者: alphaau 時間: 2008-12-22 10:13
詳情搵唔番亦諗唔起,
不過我聽過話如果要apply game theory 落marksix
佢轉波要重新計stat
佢轉放位要重新計stat
佢轉機要重新計stat
同埋, 你個stat 要包埋d 波出0黎的seq.
作者: thinkpanda 時間: 2008-12-22 12:35
原帖由 alphaau 於 2008-12-22 09:31 發表
search for john nash,
game theory
but you need more data then just the six/sever number!
How is game theory be applied to Mark Six?
作者: 阿水 時間: 2008-12-22 12:54
how can you mine from purely random thing
作者: solhen 時間: 2008-12-22 13:49
GAME THEORY (博奕理論)... 係用唔到係mark six
作者: 梁炳 時間: 2008-12-22 13:57
原帖由 阿水 於 2008-12-22 12:54 發表
how can you mine from purely random thing
will some balls(numbers) appear more times than the others
because of some extremely small difference? (weight,shape,elasticity...)
so that if we bet on those balls, we will get higher chance of winning?
作者: Tony仔 時間: 2008-12-22 14:01
原帖由 梁炳 於 2008-12-22 13:57 發表
will some balls(numbers) appear more times than the others
because of some extremely small difference? (weight,shape,elasticity...)
so that if we bet on those balls, we will get higher chance of win ...
the contribution of randomness by the factors you mentioned is far insignificant to the ball positions and velocity
[ 本帖最後由 Tony仔 於 2008-12-22 14:01 編輯 ]
作者: thinkpanda 時間: 2008-12-22 15:26
原帖由 梁炳 於 2008-12-22 13:57 發表
will some balls(numbers) appear more times than the others
because of some extremely small difference? (weight,shape,elasticity...)
so that if we bet on those balls, we will get higher chance of win ...
你咁講唔係話完全不可能, 事實上破解MD5, DES 呢類cryptographic algorithm, 都係依靠佢地得出既計算結果唔夠evenly distributed 作突破點.
但係以六合彩既情況黎講, 一年開一百次, 一百年先開到一萬次, sample size 以呢個用途黎講係太細. 未必可以mine 到一個統計上有效既結論...
作者: 阿水 時間: 2008-12-22 17:02
原帖由 thinkpanda 於 2008-12-22 03:26 PM 發表
你咁講唔係話完全不可能, 事實上破解MD5, DES 呢類cryptographic algorithm, 都係依靠佢地得出既計算結果唔夠evenly distributed 作突破點.
但係以六合彩既情況黎講, 一年開一百次, 一百年先開到一萬次, sample size 以呢個用途黎講係太細. 未必可以mine 到一個統計上有效既結論...
重有六合彩有二set 波
u don't know which set is applied
作者: xinfang 時間: 2008-12-22 17:03 標題: 回覆 8# 的帖子
Another factor to consider is, there may be more than one set of balls and machines.
It will be next to impossible to crack if no human error in the operating procedures.
(This reminds me of how Enigma is cracked...
)作者: superkufu 時間: 2008-12-22 17:41
不過實際上, 馬會又唔會甘傻比我地試到既....
其實2-3000 個sample 我都覺得ok .wor.. 當然有時要忽略multi-set of balls or multi-set of machines...
There is no way to get the 100% correct, but the nearly 100%....
作者: superkufu 時間: 2008-12-22 17:45
同埋我有覺得 所有number 既 probability 最終係nearly even.... 所以都有得諗既...
不過要想好多好多factor 去 mine 姐...
其實用naive bayes 仲好 =.=
作者: carychow 時間: 2008-12-22 21:10
我日日返工對住呢類博奕遊戲...要搵六合彩條數唔難計, 比起賭場入面玩得o既易好多.
但你搵到答案後想用黎賺錢就放棄算啦. 始終只係娛樂黎, 想中獎只有靠碰運氣.
作者: yashiharu 時間: 2008-12-23 01:01
提示: 作者被禁止或刪除 內容自動屏蔽
作者: superkufu 時間: 2008-12-23 09:10
FLASH =.= ? ...OH you are too amazing..
作者: yashiharu 時間: 2008-12-23 19:00
提示: 作者被禁止或刪除 內容自動屏蔽
作者: jj2a3 時間: 2008-12-23 20:24
原帖由 阿水 於 2008-12-22 17:02 發表
重有六合彩有二set 波
u don't know which set is applied
原來有兩set 波咖? 真係搵笨.........
作者: 阿水 時間: 2008-12-23 20:35
原帖由 jj2a3 於 2008-12-23 08:24 PM 發表
原來有兩set 波咖? 真係搵笨.........
好久之前講過
而且d 波會定期驗身
馬會對它對得非常好
