Worst-case metrics

The classification of the 12 substances is given in Table 6 with respect to the cases C (Coincidence among all four methods), S (the rank of a substance depends on the method and therefore on the formalisms to include participation. HDT, as a method based purely on the data matrix seems to be on the safe side) and case W ( Worst case, discrepancies arise because of the different approaches (non-metric (HDT) vs. metric (the other three methods)). 

FIGURE 73.2 Examples of different types of assessments that can be performed by combining performance capacity measures and reference values of different types. The upper section shows raw score values as well as statistics for a healthy normal reference population in tabular form (left). It is difficult to reach any decision by simple inspection of just the raw performance capacity values. Tabular data are used to obtain a percent normal assessment middle) and a z-score assessment right). Both of these provide a more directly interpretable result regarding subject A s impairments. The lower section shows raw score values (same as in upper section) and quantitative demands (typically worst case) imposed on the respective performance resources by task X. The lower-middle plot illustrates the process of individually assessing sufficiency of each performance resource in this task context using a threshold rule (i.e., availability must exceed demand for sufficiency). The lower-right plot illustrates a similar assessment process after computation of a stress metric for each of the performance capacities. Here, any demand that corresponds to more than a 100% stress level is obviously problematic. 

This form of verification is required for hard systems. For soft or firm components the model can be exercised to determine its average performance or other behavior metrics. The ability to combine worst-case and average-case behavior is a key aspect of the computational model. 

The error metric is a measure of a model s prediction accuracy. The software provides a number of error metrics such as squared error, worst-case error, logarithm error, median error, interquartile absolute error and signed difference for minimization. Additionally, options to maximize the correlation coefficient or the B goodness of fit or experimental hybrid that considers both absolute error and correlation are also available. Data sphtting is an important step which divides the data into a training set to generate solutions and a test set to check the accuracy of those solutions (Fig. 3.62). 

---