Yi-Chuan Lu

An ontology-based business intelligence application in a financial knowledge management system

Business intelligence (BI) applications within an enterprise range over enterprise reporting, cube and ad hoc query analysis, statistical analysis, data mining, and proactive report delivery and alerting. The most sophisticated applications of BI are statistical analysis and data mining, which involve mathematical and statistical treatment of data for correlation analysis, trend analysis, hypothesis testing, and predictive analysis. They are used by relatively small groups of users consisting of information analysts and power users, for whom data and analysis are their primary jobs. We present an ontology-based approach for BI applications, specifically in statistical analysis and data mining. We implemented our approach in financial knowledge management system (FKMS), which is able to do: (i) data extraction, transformation and loading, (ii) data cubes creation and retrieval, (iii) statistical analysis and data mining, (iv) experiment metadata management, (v) experiment retrieval for new problem solving. The resulting knowledge from each experiment defined as a knowledge set consisting of strings of data, model, parameters, and reports are stored, shared, disseminated, and thus helpful to support decision making. We finally illustrate the above claims with a process of applying data mining techniques to support corporate bonds classification.

Improved slack-based context-dependent DEA - A study of international tourist hotels in Taiwan

In this paper, we improve the slack-based measure (SBM) of efficiency in context-dependent data envelopment analysis (DEA) and apply in measuring the performance of 34 selected Taiwans International Tourist Hotels. Empirical results indicate that (1) the market differentiates five performance levels forming the benchmark structure for 34 hotels; (2) the hotels with higher attractiveness can be served as learning targets for the hotels in the lagging levels so as to establish the best path for performance improvements; (3) the hotels in the leading levels can use lower progress to analyze potential competitors in the lagging levels. The results of this study can provide hotels managers with insights of competitive advantage and help them with strategic decision making.

A visualized data analysis for bogus business entity detection

We aimed to build up a visualized data analysis process to detect potential tax evasion caused by bogus business entities. Relevant datasets of bogus business entities were extracted from large databases of 8 different systems at the Fiscal Information Agency for tax evasion pattern recognition and behavior analysis. Visualized data analysis not only helped on identifying bogus business entities, but also helped to establish a detection mechanism to discover potential tax evasion for the government.

Performance Benchmarking by Improved Slack-based Context-dependent DEA for the Hotel Industry in Taiwan

Statistics from the Taiwan Tourism Bureau (TTB) state that travelers to Taiwan in 2006 spent an average of 44.74% of their daily spending at hotels. Thus, the quality of services and the level of operational performance at hotels is key to the rise or fall of the tourism industry. Hotels in Taiwan can be divided into two groups: general and international tourist hotel (ITH). Between 1992 and 2006, the number of general tourist hotels decreased by 31%; the number of ITHs increased by 18.72%. This fact shows that large-scale ITHs have become the market mainstream; therefore, performance management of ITHs is an important research issue. A variety of previous studies dealing with this issue either adopt the comparison of financial indicators or apply data envelopment analysis (DEA) to compute relative efficiency values of decisionmaking units (DMU). Improving the performance of any inefficient DMU requires a comparison with the most efficient DMU. However, the lack of identification information concerning market competitors provides no new research perspective. The SBM context-dependent DEA proposed by Morita, Hirokawa and Zhu (2005) offers many options for performance improvement; yet it fails to shed light on how to identify potential competitors. Moreover, X has not yet certified the research method as feasible. The objectives of this research are listed as follows: (1) analyze operational performance of hotels in Taiwan, (2) provide reference paths for hotel practitioners to make gradual improvements, and (3) assist hotel practitioners in identifying potential competitors. The research issues are as follows: (1) improve the shortcoming of the research method proposed by Morita, Hirokawa and Zhu (2005), (2) establish a benchmark structure for all DMUs, and (3) perform empirical analysis of information obtained from the benchmark structure.

Research on the Elderly Living Industry Development on the Assessment Approach of Regional Economic Impact

The aging society brings changes to the living model and family structure and also triggers new industry needs and issues. Taiwans health maintenance residence develops towards the project in combination with medical system in recent years. The layout of health maintenance, customized programming hardware as well as the care services has become a trend. To this end, the public and private sectors are successively devoting resources to this market. This paper plans to use the regional computable general equilibrium model (RCGE) as the tool to make exploration on the assessment method of the impact of the development of elderly living industry on regional economy, which can be used to explain the interaction between industrial sectors in the overall economic system and can analyze the impact of economic activities in regional operation as well as the economy as a whole. In foreign literature, computable general equilibrium (CGE) discusses different regional issues such as tax, government budget and trade deficit, etc., whereas the domestic literature mainly discusses international trade, tariffs and tax reform, energy and environmental issues, etc. However, it is lack of the literature on regional computable equilibrium model by regional data. In addition to providing the reference for the public and private sectors on the elderly living industry investment and discussing the status quo of elderly living industry in New Taipei City, the study mainly focuses on the discussion of the effect of elderly living industry development by input-output model via the secondary data sorting, preparation of northern region input-output table, and proposes the estimation of investment benefit and the follow-up research proposals.

Towards Automated Optimal Equity Portfolios Discovery in a Financial Knowledge Management System

We propose a knowledge discovery and knowledge management process for equity management institutions. We realize the process with a financial knowledge management system, FKMS, that is a system platform being able to convert various sources of data into the data warehouse, to retrieve data cubes based on different power users’ commands for subsequent valuation modeling or data mining applications. We then introduced a data mining solution for equity portfolio construction using the simulated annealing algorithm. Two data sets consist of small stocks ranging from 11/86 to 10/91 and from 6/93 to 5/96 are used. The corresponding rates of return of Russell 2000 index are collected as benchmarks for evaluation based on the Sharpe ratios and the turnover ratios. The result of the simulated annealing algorithm has shown to outperform the market index as well as the gradient maximization method.

Towards Automated Optimal Equity Portfolios Discovery in a Knowledge Sharing Financial Data Warehouse

We propose a knowledge discovery process for multi-factor portfolio management on a financial decision support system. We first construct an OPen Intelligent Computing System (OPICS) to support time series management and knowledge management. A system, Cyclone, which efficiently supports financial applications, is developed under the OPICS. We then introduce a data mining solution for equity portfolio construction using the simulated annealing algorithm. Two data sets consist of small stocks ranging from 11/86 to 10/91 and from 6/93 to 5/96 are used. The corresponding rates of return of Russell 2000 index are collected as benchmarks for evaluation based on the Sharpe ratios and the turnover ratios. The result shows that the simulated annealing algorithm outperforms both the market index and the gradient maximization method.