Posted 19 December 2009 - 03:38 PM
I bought the Head First Statistics book in hopes of finding a probabilistic way to solve the proposed problem:
1-A particular product “M” has been on the market for 5 years and during this time it has been upgraded/recalled 3 times;
2-Because not all customers have sent their units to be upgraded, presently there are 4 “flavors” of this product on the market: original (M0), first upgrade (M1), second upgrade (M2) and third upgrade (M3);
3-The total number of M units on the market today is M0 + M1 + M2 + M3 = 2500;
4-The number of M units returning for service ( repair and/or upgrade ) every month is 25;
5-The flavors ( M0, M1, M2 or M3 ) of the M units coming in for service and shipped back to their owners are known;
6-What statistical tool/method/process can one use to estimate or calculate the composition of the M population today ( percentage of M0, M1, M2 and M3 ) using returning for service information as sampling data?
7-Note: the samples ( M units returning for service every month ) are biased ( not a normal distribution ) because those units are either broken, returning/recalling to fix known issues or sent by mistake.
Does anyone have any suggestion to where I should start looking for a solution?