Application of Data Mining Technique in the Performance Analysis of Shipping and Freight Enterprise and the Construction of Stock Forecast Model

Abstract

Taiwan’s economic development is closely related to Mainland China. The signing of Economic Cooperation Framework Agreement (ECFA) across the strait has great impact on Taiwan’s industry. This study is going to investigate the impact on the operation of shipping and freight related enterprises after the signing of ECFA, and the results are going to be used as reference by the government departments. First, the financial data of shipping and freight related enterprises before and after the signing of ECFA will be collected from InfoWinner Plus Database. Meanwhile, grey relational analysis and Data Envelope Analysis will be adopted to investigate whether there is significant difference between the business operation performances before and after the signing of ECFA; later on, decision tree analysis is used to investigate the major causes affecting the business operation performance; finally in this study, shipping enterprises with the best performances are selected and stock related information are collected too, then methods such as Particle Swarm Optimization optimized general regression neural network (PSO_GRNN), General Regression Neural Network (GRNN) and multiple regression are adopted respectively to set up stock forecast models to be used as reference by the public investors and the researchers. From the analysis result, it can be seen that after the signing of ECFA, the business operation performance of shipping and freight enterprises is significantly enhanced, and the forecast capability of Particle Swarm Optimization optimized general regression neural network (PSO_GRNN) model is the best.