Table 16 Preference model allowing for the order of alternatives to be the same as in the empirical ranking
From: Framework for multi-criteria assessment of classification models for the purposes of credit scoring
Criterion | Weight | Sub-criterion | Weight | Time period | Weight | Preference direction | Preference function |
|---|---|---|---|---|---|---|---|
C1-Gini | 1 | SC1.1—Training set | 1 | 2017q1 | 1 | Max | V-shaped |
SC1.2—Validation set | 2 | 2017q2 | 1.05 | ||||
C2-Accuracy | 0 | SC2.1—Training set | 0 | 2017q3 | 1.1 | ||
SC2.2—Validation set | 0 | 2017q4 | 1.15 | ||||
C3-Precision | 0 | SC3.1—Training set | 0 | 2018q1 | 1.2 | ||
SC3.2—Validation set | 0 | 2018q2 | 1.25 | ||||
C4-Recall | 0 | SC4.1—Training set | 0 | 2018q3 | 1.3 | ||
SC4.2—Validation set | 0 | 2018q4 | 1.35 | ||||
C5-F1 score | 0 | SC5.1—Training set | 0 | 2019q1 | 1.4 | ||
SC5.2—Validation set | 0 | 2019q2 | 1.45 |