BioInspired Credit Risk Analysis

Part I Credit Risk Analysis with Computational Intelligence: An Analytical Survey................................................................................. 1 1 Credit Risk Analysis with Computational Intelligence: A Review..... 3 1.1 Introduction........................................................................................ 3 1.2 Literature Collection .......................................................................... 5 1.3 Literature Investigation and Analysis ................................................ 7 1.3.1 What is Credit Risk Evaluation Problem? .................................. 8 1.3.2 Typical Techniques for Credit Risk Analysis ............................. 8 1.3.3 Comparisons of Models ............................................................ 17 1.4 Implications on Valuable Research Topics...................................... 23 1.5 Conclusions...................................................................................... 24 Part II Unitary SVM Models with Optimal Parameter Selection for Credit Risk Evaluation........................................................................................ 25 2 Credit Risk Assessment Using a Nearest-Point-Algorithm-based SVM with Design of Experiment for Parameter Selection.............. 27 2.1 Introduction...................................................................................... 27 2.2 SVM with Nearest Point Algorithm ................................................ 29 2.3 DOE-based Parameter Selection for SVM with NPA ..................... 33 2.4 Experimental Analysis..................................................................... 35 2.5 Conclusions...................................................................................... 38 3 Credit Risk Evaluation Using SVM with Direct Search for Parameter Selection .......................................................................................... 41 3.1 Introduction...................................................................................... 41 3.2 Methodology Description ................................................................ 43 3.2.1 Brief Review of LSSVM .......................................................... 43 3.2.2 Direct Search for Parameter Selection...................................... 45 3.3 Experimental Study.......................................................................... 47 3.3.1 Research Data ........................................................................... 47 XIV Table of Contents 3.3.2 Parameter Selection with Genetic Algorithm ........................... 48 3.3.3 Parameters Selection with Grid Search..................................... 49 3.3.4 Experimental Results ................................................................ 50 3.4 Conclusions......................................................................................54 Part III Hybridizing SVM and Other Computational Intelligent Techniques for Credit Risk Analysis .......................................................................... 57 4 Hybridizing Rough Sets and SVM for Credit Risk Evaluation........ 59 4.1 Introduction...................................................................................... 59 4.2 Preliminaries of Rough Sets and SVM ............................................ 61 4.2.1 Basic Concepts of Rough Sets .................................................. 61 4.2.2 Basic Ideas of Support Vector Machines.................................. 62 4.3 Proposed Hybrid Intelligent Mining System ................................... 63 4.3.1 General Framework of Hybrid Intelligent Mining System....... 63 4.3.2 2D-Reductions by Rough Sets.................................................. 64 4.3.3 Feature Selection by SVM........................................................ 65 4.3.4 Rule Generation by Rough Sets................................................ 66 4.3.5 General Procedure of the Hybrid Intelligent Mining System ... 67 4.4 Experiment Study ............................................................................ 68 4.4.1 Corporation Credit Dataset ....................................................... 69 4.4.2 Consumer Credit Dataset .......................................................... 70 4.5 Concluding Remarks........................................................................ 72 5 A Least Squares Fuzzy SVM Approach to Credit Risk Assessment 73 5.1 Introduction...................................................................................... 73 5.2 Least Squares Fuzzy SVM............................................................... 74 5.2.1 SVM.......................................................................................... 74 5.2.2 FSVM ....................................................................................... 77 5.2.3 Least Squares FSVM ................................................................ 79 5.3 Experiment Analysis........................................................................ 81 5.4 Conclusions...................................................................................... 84 6 Evaluating Credit Risk with a Bilateral-Weighted Fuzzy SVM Model.................................................................................................... 85 6.1 Introduction...................................................................................... 85 6.2 Formulation of the Bilateral-Weighted Fuzzy SVM Model ............ 89 6.2.1 Bilateral-Weighting Errors ....................................................... 89 6.2.2 Formulation Process of the Bilateral-weighted fuzzy SVM..... 91 6.2.3 Generating Membership ........................................................... 93 6.3 Empirical Analysis........................................................................... 95 Table of Contents XV 6.3.1 Dataset 1: UK Case................................................................... 96 6.3.2 Dataset 2: Japanese Case .......................................................... 98 6.3.3 Dataset 3: England Case .........................................................100 6.4 Conclusions....................................................................................102 7 Evolving Least Squares SVM for Credit Risk Analysis ..................105 7.1 Introduction....................................................................................105 7.2 SVM and LSSVM..........................................................................108 7.3 Evolving LSSVM Learning Paradigm...........................................111 7.3.1 General Framework of Evolving LSSVM Learning Method .111 7.3.2 GA-based Input Features Evolution........................................113 7.3.3 GA-based Parameters Evolution.............................................117 7.4 Research Data and Comparable Models ........................................119 7.4.1 Research Data .........................................................................119 7.4.2 Overview of Other Comparable Classification Models..........121 7.5 Experimental Results .....................................................................123 7.5.1 Empirical Analysis of GA-based Input Features Evolution....123 7.5.2 Empirical Analysis of GA-based Parameters Optimization ...126 7.5.3 Comparisons with Other Classification Models .....................129 7.6 Conclusions....................................................................................131 Part IV SVM Ensemble Learning for Credit Risk Analysis............................133 8 Credit Risk Evaluation Using a Multistage SVM Ensemble Learning Approach.............................................................................................135 8.1 Introduction....................................................................................135 8.2 Previous Studies.............................................................................138 8.3 Formulation of SVM Ensemble Learning Paradigm .....................140 8.3.1 Partitioning Original Data Set.................................................140 8.3.2 Creating Diverse Neural Network Classifiers.........................142 8.3.3 SVM Learning and Confidence Value Generation .................143 8.3.4 Selecting Appropriate Ensemble Members ............................144 8.3.5 Reliability Value Transformation ...........................................146 8.3.6 Integrating Multiple Classifiers into an Ensemble Output......146 8.4 Empirical Analysis.........................................................................148 8.4.1 Consumer Credit Risk Assessment.........................................149 8.4.2 Corporation Credit Risk Assessment ......................................151 8.5 Conclusions....................................................................................154 9 Credit Risk Analysis with a SVM-based Metamodeling Ensemble Approach.............................................................................................157 XVI Table of Contents 9.1 Introduction.................................................................................... 157 9.2 SVM-based Metamodeling Process............................................... 160 9.2.1 A Generic Metalearning Process ............................................ 160 9.2.2 An Extended Metalearning Process ........................................ 163 9.2.3 SVM-based Metamodeling Process........................................ 165 9.3 Experimental Analyses .................................................................. 173 9.3.1 Research Data and Experiment Design................................... 173 9.3.2 Experimental Results .............................................................. 174 9.4 Conclusions.................................................................................... 177 10 An Evolutionary-Programming-Based Knowledge Ensemble Model for Business Credit Risk Analysis ................................................... 179 10.1 Introduction.................................................................................. 179 10.2 EP-Based Knowledge Ensemble Methodology ........................... 181 10.2.1 Brief Introduction of Individual Data Mining Models ......... 182 10.2.2 Knowledge Ensemble based on Individual Mining Results . 185 10.3 Research Data and Experiment Design........................................ 188 10.4 Experiment Results ...................................................................... 189 10.4.1 Results of Individual Models................................................ 189 10.4.2 Identification Performance of the Knowledge Ensemble ..... 191 10.4.3 Identification Performance Comparisons ............................. 193 10.5 Conclusions.................................................................................. 195 11 An Intelligent-Agent-Based Multicriteria Fuzzy Group Decision Making Model for Credit Risk Analysis........................................ 197 11.1 Introduction.................................................................................. 197 11.2 Methodology Formulation ........................................................... 201 11.3 Experimental Study...................................................................... 206 11.3.1 An Illustrative Numerical Example ...................................... 206 11.3.2 Empirical Comparisons with Different Credit Datasets ....... 208 11.4 Conclusions and Future Directions.............................................. 221 References............................................................................................... 223 Subject Index.......................................................................................... 239 Biographies of Four Authors of the Book............................................ 243