I'm studying a manufacturing process that produces widgets, with a number of machines that breakdown or jam occasionally.
I want to figure out how many hours of studying I need to do, in order to be confident in knowing what the biggest causes of downtime are.
I will then make changes to the machines that are causing the most problems, and continue to study the process. I then want to determine how much better the process is.
How should I go about this? It seems like it should be easy, but I don't know where to start, thanks in advance!
Update #1 I think I need to start with some bounds for the problem, such as a confidence level, and population size. I'm not sure why it is the case, but people seem happy with 95% confidence. I think that the population is unknown / infinite, as it is time. So, how does my desired confidence level (95%) translate into a sample size (number of hours)?
