Statistical tools scatter plot

Analyze. Having identified the who and what of this problem, we now target the where, when, and why of the defects in the process. We use appropriate statistical analysis tools, scatter plots, SPC and SQC, Input/Output matrixes, hypothesis testing, and the like, and attempt to accurately understand what is happening in the process. 

To analyze the results of the optimization process, Kimeme provides a rich set of plots, statistical analyses and post-processing tools. Some examples of such tools are shown in Fig. 4, including for instance various scatter plots 2D and 3D, and other multi-dimensional visualization plots such as matrix and parallel plots. The typical post-processing use case involves the selection of one or more solutions from the main solution tables (for example those belonging to the Pareto front), and the choice of the desired plots, see Fig. l.b for an example. 

A final comment addresses the use of statistics. If judiciously applied, statistics is an invaluable tool for finding values of coefficients that best fit experimental data. However, caution is called for in two respects A good statistical correlation provides no guarantee that the equation or model used is indeed correct. In complex systems, the evaluation may converge on a false optimum. Also, primitive statistics programs do not distinguish between random scatter and systematic deviations. As an illustration, Figure 3.11 shows a comparison of two first-order concentration plots of data with approximately the same statistical deviation from... 

---