Ping-Tsai Chung

On data integration and data mining for developing business intelligence

Business Intelligence (BI) allows a corporations executives to acquire a better understanding of their customers, the market, supply and resources, and competitors in order to make effective strategic decisions. BI technologies provide historical, current and predictive views of business operations such as reporting, online analytical processing, business performance management, competitive intelligence, benchmarking, and predictive analytics. Web Services technologies responded quickly to help such evolution and in many situations the Web Services application is driving businesses and dictating a new way of doing business. Web information usually contains multimedia data with unstructured fashions. Through the effective analysis of companys Web information, we could make effective market analysis, compare customer feedback on similar products, discover the strengths and weaknesses of their competitors, retain highly valuable customers, and make smart business decisions. In this paper, we discuss two case studies on data integration and data mining. The first case is for the traditional data analytics using relational database techniques such as Oracle database and Cognos BI tool for integrating and mining a companys web site. The second case is for multimedia data analytics using Monago database and Pentaho BI tool for integrating and mining multimedia data presented in a companys web site. We compare both cases in aspects of Data Integration, Metadata, Query Performance and Data Analytics. Finally, we present experimental results for using the above data mining techniques and tools to better understand features of each customer group and develop customized customer reward programs.

Probabilistic Relational Database Applications for Biomedical Informatics

In the information age, information is usually communicated and exchanged with heterogeneous databases in a distributed internetworking environment. When multiple heterogeneous databases show different values for the same data item, its actual value is not known with certainty. Probabilistic relational databases have extended from the relational database model by incorporating probability measures to capture the data uncertainty associated with data items. In this paper, we apply the probabilistic relational database techniques to the area of biomedical informatics. We address the issue of database interoperability by using a data warehousing approach to resolve mismatched domains and to integrate information from different data sources. We then develop query processing strategies to effectively evaluate and answer queries from the consolidated database.

A knowledge-based decision system for healthcare diagnosis and advisory

In this paper, we present a knowledge-based decision system for healthcare. It not only performs intelligent diagnoses but also produces inferential advices for the interrelated diseases involving overweight or obese, diabetes, high blood pressure and high cholesterol conditions. Moreover, it performs deep diagnoses for the pregnant Asian women; for the unknown type of diabetes and for the risk of a heart attack and stroke increases. Also, it generates a risk report to remind patients to pay attention on their health. Our knowledge-based decision system provides the efficient and effective way to take care of the patients health, to promote the humans quality of life and to provide disease monitoring and control to alleviate or to reduce the medical condition. In the long-term, this system will help us to reduce our medicare investments and to provide high quality healthy lives.

Cluster analysis for reducing city crime rates

Data analysis plays an indispensable role in the knowledge discovery process of extracting of interesting patterns or knowledge for understanding various phenomena or wide applications. Visual Data Mining is further presenting implicit but useful knowledge from large data sets using visualization techniques, to create visual images which aid in the understanding of complex, often massive representations of data. As the amount of data managed in a database increases, the need to simplify the vast amount of data also increases. Cluster analysis is the process of classifying a large group of data items into smaller groups that share the same or similar properties. In this paper, different Clustering algorithms such as K-Means clustering, agglomerative clustering were studied and applied to the Stop, Question and Frisk Report Database, City of New York, Police Department, NYPD, for analyzing the location of the crime and stopped people using the reason of stopped in order to reduce city crime rates. Our analytic and visual results revealed that the best clustering algorithm is K-Means algorithm, and its good features ensuring that the models are helpful.