Data is expensive or in the case of time series, requires time to collect. Maximum Entropy is a shrinkage estimator that trades off bias of the parameter estimates for variance reduction. In this tutorial we will demonstrate how maximum entropy and its variants compare to ordinary least squares and ridge regression for several important problems in econometrics such as; intervention analysis, ARMA modeling, SUR models, and ill-conditioned systems. This is an introductory tutorial suitable for anyone with understanding of linear regression.

Donald James Erdman received a BS degree with High Honors in Mathematics and Chemistry from Eckerd College in St. Petersburg, Florida in 1984. He received his masters and Ph.D. in computer science from Duke University in 1986 and 1989 respectively. From 1989 to 1991 he was employed at Microelectronics Center of North Carolina, in Research Triangle Park, North Carolina developing algorithms for simulation of large differential algebraic systems. In 1991 he became a Senior Research Statistion at SAS Institute in Cary, North Carolina. His research interest include Econometrics, computational methods for large nonlinear differential algebraic systems, and estimation and simulation of nonlinear systems of equations.

Data Mining: An Overview of Process, Techniques, and Tools

Data Mining is the application of statistics, machine learning, and data visualization to business databases in order to discover actionable patterns. The resulting actions include detecting fraud, predicting market trends, identifying risk, targeting potential customers with improved accuracy, and improving relationships with existing customers. This two-hour tutorial will cover the fundamentals of data mining, including the process, techniques, and available tools. All aspects of the data mining process, including business and data understanding, data preparation, modeling, evaluation, and deployment will be discussed. An overview of Data mining techniques, including classification, prediction, association, and clustering will be included. The tutorial will conclude with a discussion of common mistakes and pitfalls to be avoided in data mining efforts.

Cheryl Howard has been active in the fields of machine learning and data mining for over 15 years. Dr. Howard is a Senior Research Scientist with Elder
Research (http://www.datamininglab.com). She has applied data mining techniques in the financial domain, resulting in improved data quality and enhanced products and services for financial service professionals. As a visiting researcher at the Institute for Applied Knowledge Processing (FAW) in Ulm, Germany, she applied adaptive machine learning to such problems as industrial quality control, medical diagnosis, and physical security. She specializes in combining non-structured data (e.g., images, acoustical signals, times series) with structured data in databases. She has published and presented papers on the application of machine learning to image analysis. Dr. Howard holds graduate degrees in Computer Science from the George Washington University and a Bachelor's degree from The University of Rochester.

Forecasters in today's telecommunications industry are required to do more with less, to forecast new products with little historical data and predict competitive outcomes. This discussion-oriented tutorial will provide helpful solutions to the problems that forecasters face every day. We will explore tips for prioritizing forecasting tasks, tricks for forecasting new products and techniques for predicting competition.

David Loomis is Chair of the ICFC conference and Assistant Professor of Economics at Illinois State University, where he teaches in the Master's program in electricity, natural gas and telecommunications economics. He is co-editor of The Future of the Telecommunications Industry: Forecasting and Demand Analysis, with Lester Taylor (Kluwer, 1999). Dr. Loomis is also the Co-Director of the Institute for Regulatory Policy Studies. Before teaching at Illinois State, Dr. Loomis worked for 12 years as a research economist at Bell Atlantic in Philadelphia focusing on issues of forecasting and demand analysis. Dr. Loomis received his Ph.D. in Economics at Temple University in Philadelphia. He received a University Outstanding Teaching Initiative Award at ISU in 1999.