Fig. 6.

Generic example of parameter reduction. In this case, it can be seen that information about parameter Z provides no benefit in describing any relationships between parameters. Instead, any significant behavior of the data can be completely described by X and Y. Thus, the system can be reduced from a 3-D parameter space to a 2-D space. Similarly, partial least squares regression or principle component analysis are used to describe multidimensional data with only the parameters that contribute the most to system behavior. Information similar to this might be used for the basic process of candidate identification or, with additional analysis, development of behavioral models.