No abstract

as sponsors of the project and particularly to Larry Wicks, contract monitor, for his many suggestions and assistance. The offering of useful ideas from all the staff of Systems Enginecring at NSWC is also greatly appreciatod. The • ~encouragement by Morley Shamblen and Alan Alexander is especially appreciated. Considerable help and encouragement was also given br several other members of the MOBA community and specifically colleagues froam the Human Engineering Laboratory at the Aberdeen Proving Ground, Don Egner. Ellsworth Shank, a"d Brenda Thein. Orbin Larson from the Navy Personnel Research and Development Center offered many suggestions and posed valuable questions concerning air operations in cities. Special thanks is also offered to Marine Corps personnel Majoru Sean Del Grosso and Mi%.dael Janay for their comments and interest. The helpful uuggestions of George Schecter, Analytics Corporation, are also acknowledged. Spx-rial appreciation is given to co-workers on the project. Thomas Kuhn and Gary spring, graduate student assistants in the Department of Geography, San Jose State UniversiLy. They worked long hours and devoted countless evenings and weekends to the project. Without their contributions in data gathering, analysis, and graphic presentation, the project would not have been possible. Finally, the views expressed in this paper are mine. For these I accept full responsibility.

While information technology has facilitated the collection of neverbefore-seen quantities of data, these data have not always provided the information needed by transportation professionals to support sound decision making. Computational intelligence (CI) has great potential to support the needs of transportation professionals. CI is a result of synergy among information processing technologies such as artificial neural networks (ANNs), fuzzy sets, and genetic algorithms. As the number of CI applications to transportation problems grows, so does the need to evaluate these systems. The issue of validating and evaluating transportation CI applications is addressed. A case study that evaluates the effectiveness of two CI paradigms, case-based reasoning and ANNs, for estimating the benefits of real-time traffic diversion is presented. The case study illustrates the need for regarding validation and evaluation as a part of the development effort and the need for tuning the design parameters of CI paradigms.

Expert system validation—that is, testing systems to ascertain whether they achieve acceptable performance levels—has with few exceptions been ad hoc, informal, and of dubious value. Very few efforts have been made in this regard in the transportation area. A discussion of the major issues involved in validating expert systems is provided, as is a review of the work that has been done in this area. The review includes a definition of validation within the context of the overall evaluation process, descriptions and critiques of several approaches to validation, and descriptions of guidelines that have been developed for this purpose.

No abstract