Introduction to Process Modeling
What are some of the different statistical methods for model building?
Selecting an Appropriate Stat Method: General Case
For many types of data analysis problems there are no more than a couple of
general approaches to be considered on the route to the problem's solution.
For example, there is often a dichotomy between highly-efficient methods
appropriate for data with noise from a normal distribution and more general
methods for data with other types of noise. Within the different approaches
for a specific problem type, there are usually at most a few competing
statistical tools that can be used to obtain an appropriate solution. The
bottom line for most types of data analysis problems is that selection of the
best statistical method to solve the problem is largely determined by
the goal of the analysis and the nature of the data.
Selecting an Appropriate Stat Method: Modeling
Model building, however, is different from most other areas of statistics
with regard to method selection. There are more general approaches and more
competing techniques available for model building than for most other types of
problems. There is often more than one statistical tool that can be effectively
applied to a given modeling application. The large menu of methods applicable
to modeling problems means that there is both more opportunity for effective and
efficient solutions and more potential to spend time doing different analyses,
comparing different solutions and mastering the use of different tools. The
remainder of this section will introduce and briefly discuss some of the
most popular and well-established statistical techniques that are useful for
different model building situations.
Process Modeling Methods
- Linear Least Squares Regression
- Nonlinear Least Squares Regression
- Weighted Least Squares Regression
- LOESS (aka LOWESS)