Statistics and six sigma?

I don't think this requires statistics just yet. I think you need to find each error and track back to the original cause. Once you start getting a few root causes you can start grouping them subjectively (you might want to double check with someone else to ensure that your essment is the same as others). Once you have the groups you can simply find those that account for 80% or so of errors and focus on them first. Once that's done, look at the next 80% and so on.