首页--工业技术论文--自动化技术、计算机技术论文--计算技术、计算机技术论文--计算机软件论文--程序设计、软件工程论文--程序设计论文

An Expectation Maximization Application for Decision Tree Classifiers on Datasets with Missing Values

Abstract第4-5页
摘要第6-10页
Table of Contents第10-12页
List of Abbreviations第12-13页
List of Figures第13-14页
List of Tables第14-15页
Chapter 1 Introduction第15-19页
    1.1 Motivation第15页
    1.2 Research Objectives第15页
    1.3 Introduction to Data Classification第15-16页
    1.4. Classification Techniques第16-17页
    1.5 Introduction to the Missingness of Data第17页
    1.6 Expectation Maximization Algorithm第17页
    1.7 Structure of the Thesis第17-19页
Chapter 2 Data Missingness and Supervised Classification Techniques第19-28页
    2.1 Missing Completely At Random(MCAR)第19页
    2.2 Missing At Random(MAR)第19-20页
    2.3 Missing Not at Random(MNAR)第20-21页
    2.4 Supervised Classification Techniques第21-28页
        2.4.1 Naive Bayes Classification第21-23页
        2.4.2 Decision Trees第23-27页
            2.4.2.1 Steps in Mining with Decision Trees第24-25页
            2.4.2.2 Attribute Selection第25-26页
            2.4.2.3 Advantages of Decision Trees第26页
            2.4.2.4 Applications of Decision Trees第26-27页
        2.4.3 Lazy Learning Classification第27-28页
            2.4.3.1 Introduction to Lazy Learning第27页
            2.4.3.2 Instance-based typical approaches第27-28页
Chapter 3 EM-Based Bayesian Network Imputation第28-38页
    3.1 Problem Statement第28-29页
    3.2 Data Imputation Methods第29-30页
        3.2.1 Mean Imputation第29-30页
        3.2.2 Hot Deck Imputation第30页
        3.2.3 Cold Deck Imputation第30页
        3.2.4 K -Nearest Neighbour Imputation第30页
    3.3 The Expectation Maximization algorithm第30-31页
    3.4 Theory of Our Algorithm第31-33页
    3.5. Our Expectation Maximization Bayesian Network Imputation Algorithm第33-38页
        3.5.1 Introduction第33-34页
        3.5.2 Symbol Definition第34-35页
        3.5.3 Algorithm Description and Analysis第35-36页
        3.5.4 Pseudocode of EBN Imputation Algorithm第36-38页
Chapter 4:Experimental Design and Analysis of Results第38-48页
    4.1 Implementing the EBN Algorithm第38页
    4.2. Methodological experimental approach第38-39页
    4.3 Selection and Setup of Training and Testing Datasets第39-40页
        4.3.1 Choosing UCI Repository Datasets and other Datasets第39-40页
        4.3.2 Introduction of different missing rates第40页
    4.4 EBN Evaluation第40-46页
        4.4.1 Evaluating EBN using the Four Original Datasets第41-43页
        4.4.2 Evaluating EBN using Different Missingness of Data Rates第43-46页
    4.5 Discussions第46-48页
Chapter 5 Conclusion and Future Work第48-50页
    5.1 Conclusion and Contribution第48-49页
    5.2 Future Work第49-50页
Acknowledgment第50-51页
References第51-56页
Appendix:List of publication and achievements第56页

论文共56页,点击 下载论文
上一篇:巩义市涉村镇旅游业发展问题研究
下一篇:六阶各向同性Descartes张量