Input: 1. Target dataset ; 2) Response variable CU; 3) The number of respondents N; |
Output: 1. Model selected Mk; 2. The metrics ; 3) Knowledge representation: CU |
1. Start |
2. for do |
3. Compute ROC Values for each model |
4. Assign weight for each model based on user’s requirements |
5. Compute Confusion matrix for each model |
6. Assign weight for each model based on user’s requirements |
7. Test the effect of Data Imbalance problem for each model |
8. Test Statistical Significance using F-test |
9. Practicability and Applicability: Identify as yes or no |
10. Simplicity of Model Interpretation: Identify as yes or no |
11. Established knowledge: Identify as Positive, Negative, Neutral, Unstudied |
12. Computational cost: identify as High, Moderate, and Low |
13. Compute the hybrid multidimensional metrics using equation (4) |
14. Find s.t |
15. end for |
16. return Mk; ; CU |