Software Process Models

  If by process model you mean a correlation equation built to relate process inputs to process outputs then the act of constructing such a model should result in an equation that suggests relations between the trending of your inputs and your final outputs over the ranges of the input variables.  Usually one measures the “effectiveness” of such models by using the model to predict a given output for a given set of inputs. If the model “works” then what should happen is once you have set up your process in accordance with the levels of the input variables of the model, the process output should fall within the prediction limits associated with the predicted output.
  Model construction and model selection is the province of regression methods and physical understanding of the process.  The broad brush picture is this:
  1. Get some representative data
  2. Record the inputs and outputs
  3. Run a regression to develop a model or models
  4. Use your physical understanding of the process to make the final selection of the model
  5. Test the model as described in the opening paragraph.
   The above is an extreme simplification of the process.  Each one of the steps outlined above requires a lot of work and understanding.  The issues of proper data gathering, model building, and model selection are all subjects of multiple semester courses at the college level.  If you have no experience with any of this you will need to do some reading.  For starters, I’d recommend the following:
1. The Cartoon Guide to Statistics – Gonick and Smith Chapters 2, 6, 10 and 11.  This will introduce you to some of the issues surrounding your questions.
2. Applied Regression Analysis – Draper and Smith – Chapters 1,3,4,5,6  as a minimum.  Chapter 2 is a restatement of Chapter 1 in matrix terms.  These chapters will give you some understanding of the methods and issues of regression.
3. Statistics for Experimenters – Box, Hunter, and Hunter – Chapters 1,2,9,10,11. These chapters will introduce you to the issues of data gathering.