Time Series Prediction Using Echo State Network

发布时间：2023-05-24 22:07

　　

【文章页数】：71 页

【学位级别】：硕士

【文章目录】：
Abstract
Chapter 1 Introduction
    1.1 Research background and significance
    1.2 Research status at home and abroad
        1.2.1 Stock price prediction research in abroad
        1.2.2 Current status of domestic research on the stock price prediction
    1.3 The research contents and chapters arrangement of this thesis
        1.3.1 Research content
        1.3.2 Primary algorithm selection and proposal of the general model
        1.3.3 Short-term stock price prediction based on ESN Mackey-Glass model
        1.3.4 Arrangement of chapters in this thesis
Chapter 2 Basic Principles of The ESN Network
    2.1 Initial Idea
    2.2 The Input Data
    2.3 Characteristics of the Echo State Network and Reservoir
    2.4 Training and The Output Data
    2.5 About the Reservoir
        2.5.1 The Size of the Reservoir
        2.5.2 Spectral Radius
        2.5.3 Input Scaling
        2.5.4 Leaking Rate
    2.6 Summary
Chapter 3 Method Development
    3.1 Mackey-Glass System
    3.2 Setting Constant Bias
        3.2.1 The preparation of Input Data
        3.2.2 Construction of the Reservoir
        3.2.3 The Training & Prediction Formula
        3.2.4 Feedback & Result
    3.3 Spectral radius, Leaking Rate and Reservoir size
        3.3.1 Gradient descent
        3.3.2 Grid Search
        3.3.3 The Result
    3.4 Summary
Chapter 4 Stock Market prediction based on ESN
    4.0 Nonlinear Characteristics
    4.1 S&P 500 Data
        4.1.1 Processing of the Data
    4.2 Benchmarking
    4.3 Parameter Optimization
    4.4 The Reservoir
    4.5 Results
    4.6 The committee method
Chapter 5 Analysis of Results & Conclusion
    5.1 Result Analysis
    5.2 Conclusion
References
Acknowledgement
附件



本文编号：3822370