I’m looking to try and see if I can create a predictive model for labor standards. Our shop collects labor at the job level instead of the piece part level. I would love to change this, but it’s a bit of old dogs new tricks issue. So what I have is basic labor categories (weld, fab, assembly etc.), which are aggregated by job. Each of the Jobs has the full BOM and Operations on it, and the operations are mass completed simply to make backflushing work. So I have factors like, count of operations, count of parts related to operation, and weight of the parts. Actual labor is collected at assembly 0 with a set of dummy operations just for the sake of collecting labor hours for the job.
What I would like to do is find a way to feed this aggregated data along with as many factors as would be available for a new BOM into some sort of process that can analyze the factors to determine for each factor how much if any of a overall effect that factor has on the total labor. (you know, like the stuff you see in sociological studies where the likely hood of a person doing x is increased by y percent when their parents own a zoo or something like that). With that information, I would like to try to create a formula(s) that could be applied to create the labor standards populated in the MOMs.
I’m not very strong in the statistics that is needed to do this. I downloaded r-project to try and see what it could do, and while it looks pretty powerful, I do not know nearly enough do anything more powerful than I could do in excel. It’s really more of a programming language than UI. I have some very basic things already done in excel, but it was a very manual fit over a limited set of jobs.
Basically, the limitation really isn’t the software, it’s my knowledge of statistics. I’m wondering if anyone knows of a software or website that can be fed tabular data and fit some sort of regressions for more than just 2 dimensional pairs, (x,y), and does a good job of guiding the user on how to use it.