A Web-based Computer Program for Determining Group Classification, page 3 of 4
In Example 2, the means of the two distributions differ by two units, and the populations have the same variability and base rates. However, the utility of correctly identifying a Group 2 case as a Group 2 case is relatively very high, with a utility estimate of 10. There is one cut-point of .85. Classifying cases with scores less than .85 as Group 2 yields 97% correct classification for Group 2 and 56% correct classification accuracy for Group 1. Because of the high utility of correctly classifying Group 2 individuals, it’s in one’s interest to be more willing to classify cases as Group 2, even if this approach allows more of the less costly Group 1 misclassifications.
Example 2. One cut-point.
530 total views, 2 views today