Introduction to Advanced Analytics Services for Managerial Decision Support Minitrack

Abstract

This minitrack consists of ten papers involving theory and practice of service-based analytics (i.e., data, text and Web mining) for support if managerial decision making. These ten papers illustrate a diverse set of approaches, demonstrating the variety of ways in which modern information technologies can be applied to today’s complex decision situations. The papers in this minitrack offer some insight into the efforts to more effectively and efficiently use information technology tools to extract knowledge and better understand our rapidly changing world. The ten papers are grouped into three topic areas (as three sessions): (1) Text mining application, (2) Data mining applications, and (3) Advanced analytics applications. In the first group (text mining applications) we have two papers that apply unique capabilities of text mining into understanding and analyzing financial markets. The paper by Siering proposes using text mining and its derivative, sentiment analysis, to support intraday investment decisions, while the paper by Hagenau et al. proposes prediction of stock prices based on automatically “reading” financial news by using context-specific features. The third paper in this group (by Napoletano et al.) proposes use of a novel method “mixed graph of terms” as opposed traditional “bags of words” representation of a text in document retrieval and making better sense out of unstructured/textual data sources. In the second group (data mining applications) we have two papers that apply knowledge discovery techniques to medical datasets. The paper by Zurada talks about a study where they applied seven different prediction algorithms to predict the risk of low back disorders. In this study Zurada claims to have proposed a more systematic and reliable approach to creating and validating classifiers to better distinguish between low and high risk manual lifting jobs that contribute to low back disorders. The second paper summarizes a data-and-text mining study where by Erraguntla et al developed algorithms to handle missing ICD 9 codes in medical datasets. Their approach involved developing a prediction model for the ICD 9 codes based on other associated attributes like medical diagnosis, medical remarks, and patient statements. They used text mining methods to extract key concepts from textual patient records, and used nearest neighborhood based classification algorithms to predict the missing ICD 9 codes. The third paper in this group (by Kim at al) proposes an ensemble model, which is based on multiple SVM classifiers, to address churner identification problems in the mobile telecommunication industry, a sector in which the role of customer retention has become increasingly important due to its very competitive business environment. According to their comparison results, the performance of the ensemble model was superior to all single and ensemble models. In the third group we have four very diverse studies. Paper by Seng and Ling proposes a pool-based cost sensitive active learning framework that requires fewer number of examples yet produces a smaller total cost compared to the previous methods. Paper by Mair et al reports on an interesting empirical study of software projects managers using a case-based reasoning tool. Their aim was to explore the interaction of cognitive processes and personality of software project managers undertaking tool-supported estimation tasks such as effort and cost prediction. The paper by Soper et al reports on a study where they mined institutional identities using n-grams (a text mining technique). They demonstrated the utility of their n-gram analysis tool in revealing identity of an academic journal, namely Communications of the ACM. The last paper in this group (by Nuhn et al) was about using clustering methods for the processing of the complex landslide simulation results to support decision making and learning. 2012 45th Hawaii International Conference on System Sciences