本研究對於金融網絡特徵開發一個分析工具，我們應用R及Shiny開發的統計網絡分析軟體，簡稱SNARS，可以用來分析網絡資料和視覺化網絡的特徵。本軟體包含兩個部分。第一部分，用來探討金融網路的特徵，包含網絡的節點數、連結數、密度、自由度、自由度分佈、直徑以及遞移性。使用者可藉由這些特徵來了解金融網絡的動態結構。第二部分，我們將這些網絡特徵視覺化，分別以靜態和動態的方式呈現，並建立視覺化的界面，幫助使用者探討網絡的性質。最後我們藉由實證資料說明使用SNARS工具的方法，及如何偵測網絡中扮演重要地位的金融機構。我們採用的資料包括美國的銀行、經紀商以及保險公司，共七十一家金融機構的對數報酬率，採用的金融網絡模型則是使用向量自我迴歸及LASSO正規化準則所建立。

In this study we developed an R+Shiny tool for financial network, called SNARS,
which can be utilized to analyze and visualize the characteristics of networks. The SNARS
tool includes two parts. The first part provide tools for investigating characteristics of
financial network, including nodes, edges, density, degree, degree distribution, diameter
and transitivity. We utilize the characteristics to understand the structure of financial
networks. In the second part, we develop a user friendly visualization interface which
can be used to visualize the network characteristics both in static and dynamic ways. For
illustration, we consider the financial network constructed by Lasso penalized Vector Auto-
Regression (LVAR) model. Log returns of seventy one stocks, which include banks, brokers
and insurance companies in the USA, are used to build the LVAR model. We use SNARS
to detect financial institutes which play important roles in the network.

論文審定書 i
致謝 ii
摘要Š iii
Abstract iv
1 Introduction 1
2 Network Characteristics 2
3 SNARS description 6
3.1 Data import . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3.2 Simulation data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
3.3 Data visualization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.3.1 Visualization of topic 1 . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.3.2 Visualization of topic 2 . . . . . . . . . . . . . . . . . . . . . . . . . 17
4 Financial network example 26
4.1 Data Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
4.2 Lasso VAR Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
4.3 Data visualization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
4.3.1 Visualization of topic 1 . . . . . . . . . . . . . . . . . . . . . . . . . 31
4.3.2 Visualization of topic 2 . . . . . . . . . . . . . . . . . . . . . . . . . 36
5 Conclusion 47
References 49
Appendix 50

Basu, S., Shojaie, A. and Michailidis, G. (2015). Network granger causality with inherent
grouping structure. Journal of Machine Learning Research, 5, 417-453.
Clauset, A., Shalizi, C.R. and Newman, M.E.J. (2009). Power-law distributions in
empirical data. SIAM Review, 51, 661-703.
Yang, B.C. (2017). A two-stage nancial network model. Master Thesis, Department
of Applied Mathematics, National Sun Yat-sen University.
Yoav Benjamini and Yosef Hochberg. (1995). Controlling the False Discovery Rate: A
Practical and Powerful Approach to Multiple Testing. Journal of the Royal Statistical
Society. Series B (Methodological), 57, 289-300.
Complexitylabs, "Network Degree Distribution," in Complexity Labs | Com-
plex Systems & Systems Thinking, January 6, 2016 http://complexitylabs.io/
degree-distribution/.
The Icons. http://fontawesome.io/icons/.