| Summary: | 碩士 === 國立臺東大學 === 資訊管理學系碩士班 === 99 === Defense budget in the central government budget accounts for about 15-19% ,which approximately amounts to more than 320 billion NT dollars. Possessing a huge amount of funds, the Ministry of National Defense is usually the focus of attention by all the people in taiwan. Therefore, the performance of business execution and quality of management of the financial unit responsible for defense budget payment definitely affects the operation of financial business in the Ministry of National Defense.
To understand the practical situation of financial business development of the ministry, and to assist management class to master the general situation of operation in the unit thoroughly, in order to strengthen correction of weakness and to improve competitive advantage of the organization, the thesis seeks variables and non-variables affecting performance of the unit and constructs early warning model possessing efficiency of prediction and identification by the annual audit system of the financial center. The thesis hopes to detect potential crises beforehand.
General private enterprises have done research for years and preformed excellently on application and development of financial early warning model. This thesis constructs early warning model applied to the military unit by logistic regression and artificial neural network analysis theory in the data from 2005 to 2010. The empirical results show that salary distribution, centralized payment, information processing and internal management variables of the internal audit item are the key variables to determine whether the unit works regularly or not. Moreover, the early warning model of backpropagation artificial neural network, constructed by the variables mentioned before, provides a better predictive ability than that of logistic regression model, regardless of training samples or Test samples. Its accuracy rate is up to 93% to 95%.
|
|---|
