本帖最後由 dye 於 2010/9/2 00:39 編輯
Read again
----------------
如果你24個橙都檢查一下，的確只有4%左右爛。但唔爛唔代表好。
而且，有人檢查過所有神職人員嗎?
例如，一箱24個橙，你求其拿1個檢查，發現係爛既。你可以話其他23個唔爛咩?你只能話至少4%左右係爛。
又或，一箱24個橙，你求其拿10個檢查，發現1個係爛既，你可以話90%唔爛咩?呢個問題就睇你要有幾多把握。當然，你仍能話至少4%左右係爛。
下次買燈用下腦喇，咪比人滾。
----------------
Sample size (relative to population) matters!
You just failed your basic statistic course again!
The conclusion is neither they are all rotten, nor all good, nor even 4% rotten.
---------------
In real life shopping, buying 4%-8% rate of rotten orange is simply not acceptable. If you already know at least one of the orange is rotten in the basket, why are you still buying the particular basket? Unless the oranges are packaged, you are expected to examine each of them before buying. (I guess you never go to supermarket.)
In a commercial sense, operating profit is commonly set around 3-5% (see Li and Fung annual report on 2009, for 3% example, 蒙牛乳業, for 5% example), having a 4%-8% rate of rotten orange is enough to make a loss.
If you are conducting professional business, your rate of error should be aiming at below 0.5%. If you are on typical industrial business, you rate of error is aimed at below 3%. |
2010/8/24 18:47
| 
2010/9/8 21:39
| 