Research On The Network Of Bank Payment Flows Based On Networked Data Mining

Today financial innovation continues to accelerate, the financial system has become increasingly complex. All kinds of financial products, financial tools, the global financial markets, financial institutions and even agencies besides other financial institutions are intricately linked together. This means that the modern financial system of constituent nodes has linked. This link can be shown by the form of a complex network. We call it the financial network.In2008, financial crisis led the world economy to suffer heavy losses. The generation of the financial crisis causes us to consider the safety of the entire financial system. The studies had shown that use the micro-level management is difficult to achieve the stability of the entire financial system. It needs the new macro-regulation and systemic risk management to explore methods. After the financial crisis, one of the core content of the international financial system reform is to vigorously promote the implementation of macro-prudential supervision of the financial system. We combine the economic activity, the behavior of financial markets and financial institutions as a whole to consider. Use a global point to evaluate and guard against the risk. ECB Financial Stability Review (2010) pointed out. The link between modern financial characteristics and the complex network analysis methods needs to be explored.From biological neural networks, telecommunications and the electric power network to the Internet, These nodes are huge and the link relationships are complex. We call them complex network. Because of the link, the changes in the adjacent nodes have a major impact. Therefore, we cannot analyze the characteristics and behavior of one or some of the nodes alone. We need explore the node as a whole.The traditional data mining is to find the feature which is hidden in the massive data. The complex network theory is also found link pattern among the nodes in the network. Therefore, the way to analyze the complex systems with complex network theory similar to the methods used the data mining.Financial network belong to a complex networks which is a new way to explore the financial network with data mining methods. Its advantages are analyzing the system from an overall perspective with combined data mining technology, not to analysis a node. This will lead to a better understanding of the financial system. Prediction and assessment of financial risk and the rational distribution of financial infrastructure are of great significance.Bank belongs to major component of the financial system, it is very important for a country’s financial system. Many scholars had made a lot of research on the banking network with different perspective. Using data of HVPS to construct a network of inter-bank flow is more timely and accurate to explore the bank network.This article uses network data mining techniques to explore the data of inter-bank payment flow. Build a model of weighted complex network topology to analyze the statistical characteristics of the topology. The analysis of network structure has a lower clustering coefficient. According to the network community discovery, the five state-owned commercial bank formed a stable community.This paper studied the visualization of the payment flows networks, the topological properties of the network and structure measures. It is concluded into five parts as followed.Section one use the complex network to explore the bank network. At first, it introduces the comments of the complex network theory, including the basic concepts of complex networks, the statistical features and the applications in the banking network. At the same time, I summarize the achievements of the present study in the bank network.Section two explores the theory of network data mining. Combine the traditional data mining theory with the complex network way. I introduce the theory of traditional data mining, pattern and mining process. Compare and summarize the difference between the network data mining and traditional data mining.Section three explores the data and its visualization. Describe the data of the inter-bank payment flow first. Introduce the payment hierarchy, payment systems and HVPS of China. Then do a briefly analysis to the overall characteristics. Second, describe the visualization of complex networks, including the rise of visual technology, the development and the visualization algorithms.Section four analyzes the capital flow network topological properties and the structural measure. The first phase focuses on the statistical features of the two networks, the edge weights and node strength. First, describe the edge weights and strengths of the distribution characteristics and estimate the power exponent. The second part of the network structure measure includes two parts:studies the assortative mixing and clustering coefficient of the networks. It also assesses the effect of threshold to the correlation and clustering coefficient.Section five explores the inter-bank network structure and finds the community of network. Use the two community’s discovery algorithms:spectral clustering algorithm and the overlap of hierarchy algorithm,which divided the banks node into several different communities.There are two innovation points.One is to combine the idea of the traditional data mining techniques with complex network theory to explore a complex network. The other is to use the community finding algorithm to divide the inter-bank payment flow network into different communities.