Fluff on top of the fluff

Contents

Abstract in Thai

Abstract in English 

Acknowledgement    

Contentsviii

List of Figures

List of Tables

CHAPTER 
1   INTRODUCTION
      1.1   Motivation and Problem Description
      1.2   The Objective of Research
      1.3   The Scope of Study
2   BACKGROUND, METHODLOGY AND MODEL DESIGN 
      2.1   Industry Background and Data Selection
       2.1.1   Industry Background
       2.1.2   Data Selection
      2.2   Decision tree Modeling  
       2.2.1   Recursive Partitioning 
       2.2.2   C45 Algorithm  
      2.3   MLP Neural Network Modeling  
       2.3.1   Introduction to Neural Networks 
       2.3.2   MLP Neural Network Architecture 
       2.3.3   Learning Algorithms 
        2.3.3.1   Levenberg-Marquardt Algorithm  
      2.4   Dimension Reducing Methods 
       2.4.1   Principal Components Analysis 
      2.5   Prediction Process Flow Design 
3   EXPERIMENTAL APPLICATION 
      3.1   Data Physical Preprocessing 
      3.2   Input Pruning 
       3.2.1   Data Normalization 
       3.2.2   Input Reduction 
      3.3   Tuning Neural Networks Models 
       3.3.1   Data Partition 
       3.3.2   Preliminary Training 
       3.3.3   Early Stop 
       3.3.4   Direct Connections 
       3.3.5   Number of Hidden Units 
4   EXPERIMENT RESULT  
      4.1   Experimental processing results 
      4.2   Accuracy Evaluation 
5   CONCLUSIONS AND FUTURE WORKS 
      5.1   Conclusions 
      5.2   Discussions 
      5.3   Future works  
REFERENCES 
APPENDICES 
     Appendix A Telecommunication Business Architecture SAS solution (TIS) 
     Appendix B Coding 
VITAE