Efficiency Classification by integrating Naïve Bayesian and Dynamic Bayesian Networks.

• Main Authors: Bo-Shan Chen, 陳柏杉
• Other Authors: Han-Ying Kao
• Format: Others
• Published: 2012
• Online Access: http://ndltd.ncl.edu.tw/handle/75gqax

碩士 === 國立東華大學 === 資訊工程學系 === 100 === In contrast with business, nonprofit organizations have specific missions overtopping profitability. For example, research institutions aim at research and development (R&D) activities and achievements. However, financial performance usually plays an important role in their organizational missions. Therefore in evaluating these institutions operational as well as financial performance must be considered. This study develops a hybrid approach in multi-period efficiency classification by integrating Data Envelopment Analysis (DEA), Naïve Bayesian Networks (NBN) and Dynamic Bayesian Networks (DBN). There are three components in this work. First, the efficiencies of the decision making units (DMUs) are estimated by DEA for a series of time. Second, the intra-period naïve Bayesian network is proposed for classifying the efficiencies for the DMUs. Third, the DBNs are learned in formulating the inter-period dependencies between the efficiencies. In training the DBN with the Markovian property, the temporal dependency is assumed unchanged over time. However, in data-driven parameter learning, it is difficult to identify the stationary temporal relationships in DBN. So we introduce the fuzzy parameters for incorporating the variation of the dynamic dependency. We conduct a case study of higher education performance in Taiwan to demonstrate the usability of this design.