Forecast techniques in production environment


In order to estimate/forecast the production time on new or slighly changed products we are creating a model.
The goal of the model is to simply fill in a few parameters to be able to estimate the total production time.

In short; we produce laminate so the variables are either cutting/sawing the laminate, followed by the amount of rows & clicks is needed to produce a sample board.

I have gathered the following data based over 3 workcenters

SAWING (important to note they also need to open the box of laminate, which is a 'fixed' set-up time, but also highly variable)
- total production time
- Quantity (boxes) sawed during this production
- Number of saw cuts per box

- total production time
- amount of rows
- amount of clicks

- total production time
- amount of cuts needed

What are the best statistical ways in order estimate future production timings?

I have already tried using 'regression' under Data Analysis in Excel, and putting the intercept sometimes on zero in order to create beta's.
I have not yet tried to perform calculations with my date for example logs, ², sqrt,...

Any suggestions/tips?


No cake for spunky
There are two approaches here. One is cross sectional, regression being the most obvious. The other assumes its a form of time series, where ARIMA with external regressors is an option.

Personally I am not sure you are predicting the future here, what you are predicting is what levels of Y there will be if you have a certain X. You are only really doing a time series projection if you assume this relationship will change over time or events in the past influence it.