sample size determination

#1
Sample size determination:
Our database: There is a historical code preference from our customer. We like to give some credit to the customer who fulfill certain condition and above our cut-off point. Ex. Customer shows up the database 20 times, but only fulfils the condition 5 times ( 25%), This is less than our cut-off (70%). There fore it will be rejected. But if customer shows up say 20 times and fulfills the condition 15 times ( 75%) this will be accepted.

Here how the data looks.


Code n N percent
GGGG 10 20 50%
YYYY 1 10 10%
KKKK 15 20 75%
PPPP 44 300 15%
HHHH 9 9 100%
FFFF 123 123 100%
BBBB 22 234 9%
…..

….
ZZZZ 4 4 100%

Where code = historical customer preference
n = number of times customer fulfills condition
N = total frequency of the customer


My question is : Is there is way ( statistically) we can determine the sample size. Meaning to say the number of times fulfils the condition has to be greater than ex. 10 (n>10) in order to qualify our analysis, and any thing less than 10 will atomically remove our observation ex . code YYYY & ZZZZ will not qualify if we choose sample size less than 10.

How do you determine the n size?


Hope to get some help on this.

Thanks.

Jimmy.