From: Framework for multi-criteria assessment of classification models for the purposes of credit scoring
Purpose and subject of the study | No. of classifiers (alternatives) | No. of criteria | Data sets | Applied MCDM methods | Refs. |
---|---|---|---|---|---|
Use of a set of MCDM methods to evaluate classification algorithms for software defect detection | 38 | 13 | 10 public-domain software defect datasets | DEA, TOPSIS, ELECTRE and PROMETHEE II | [16] |
An approach to resolve disagreements among MCDM methods based on Spearman’s rank correlation coefficient | 17 | 10 | over 11 public-domain binary classification datasets | TOPSIS, ELECTRE, GRA, VIKOR, PROMETHEE | [54] |
The choice of classification algorithm in Machine Learning | 7 | 10 | Australian public domain credit data set | FAHP, TOPSIS, SAW | [55] |
Finding of a robust classifier, which is suitable for consideration as the base learner, while designing a host-based or network-based intrusion detection system | 54 | 16 | the NSLKDD, ISCXIDS2012, CICIDS2017 datasets | TOPSIS | [56] |
An accurate multi-criteria decision making methodology (AMD) which empirically evaluates and ranks classifiers’ and allow end users or experts to choose the top ranked classifier for their applications AMD methodology presents an expert group-based criteria selection method | 35 | 4 (selected by experts out of 8 features) | 15 publicly available UCI and OpenML datasets | AHP, TOPSIS | [57] |
Comparing the performance of algorithms those are used to predict diabetes using data mining techniques | 5 | 3 | 1 data set from UCI machine learning data repository | comparison of criterion values | [58] |
A new classification algorithm recommendation method based on link prediction between data sets and classification algorithms | 21 | 5 | 131 publicly available UCI data sets | proposition of own method based on: prediction and Data and Algorithm Relationship (DAR) Network | [59] |