Louis A. Le Blanc

A hybrid intelligent system for predicting bank holding structures

A composite model of neural network and rough sets components was constructed to predict a sample of bank holding patterns. The final model was able to correctly classify 96% of a testing set of four types of bank holding structures. Holding structure is defined as the number of banks under common ownership. For this study, forms of bank holding structure include: banks that are not owned by another company, single banks that are held by another firm, pairs of banks that are held by another enterprise, and three or more banks that are held by another company. Initially, input to the neural network model was 28 financial ratios for more than 200 banks in Arkansas for 1992. The 28 ratios are organized by categories such as liquidity, credit risk, leverage, efficiency, and profitability. The ratios were constructed with 70 bank variables such as net worth, deposits, total assets, net loans, total operating income, etc. The first neural network model correctly classified 84% of the testing set at a tolerance level of 0.20. Another artificial intelligence (AI) procedure known as two-dimensional rough sets was then applied to the dataset. Rough sets reduced the number of input variables from 28 to 18, a drop of 36% in the number of input variables. This version of rough sets also eliminated a number of records, thereby reducing the information system (i.e., matrix) on both vertical and horizontal dimensions. A second neural network was trained with the reduced number of input variables and records. This network correctly classified 96% of the testing set at a tolerance level of 0.20, an increase of 11% in the accuracy of the prediction. By applying two-dimensional reducts to the dataset of financial ratios, the predictive accuracy of the neural network model was improved substantially. Banking institutions that are prime candidates for mergers or acquisitions can then be more accurately identified through the use of this hybrid decision support system (DSS) which combines different types of AI techniques for the purposes of data management and modeling.

A neural network for transportation safety modeling

Abstract Accidents serve as an operational measure of marine safety, and specifically the safety of vessels, crews, and cargoes. The ability to accurately predict the type of vessel accident with such input variables as time, location, weather, river stage, and traffic could significantly reduce marine casualties by alerting port authorities and navigation groups as to the likelihood of a specific kind of casualty. In this paper, three models were developed to predict vessel accidents on the lower Mississippi River. These models are a neural network, multiple discriminant analysis and logistic regression. The predictive capability for vessel accidents of a neural network is compared with multiple discriminant analysis and logistic regression. The percent of “grouped” cases correctly classified is 80% (36 of the 45 cases in the testing set) for the neural network, if nonclassified cases are treated as incorrectly classified by neural network. The percent of “grouped” cases correctly classified by this network is 90% (36 of 40 cases) if nonclassified cases are excluded from the calculation. Discriminant analysis and logistic regression were able to correctly classify only 53% and 56% respectively, of accident cases into three casualty groups: collisions, rammings, or groundings.

An evaluation and selection methodology for expert system shells

Abstract This article illustrates an evaluation and selection methodology for expert system (ES) shells. The methodology incorporates three stages: (1) ES shell screaning, (2) shell evaluation, and (3) assurance of final ES shell selection. Initially, developing a short list through screening of commercial shell products determines whether appropriate software exists and narrows the field of available expert system software products for detailed consideration. The second stage determines which of the remaining ES shells (the finalists) best meets the needs of the organization, from both functional and technical perspectives. The final stage compares user requirements with the features of the selected ES software by defining how these requirements will be satisfied by building expert system applications with the selected product. The methodology also considers the possibility that, at any stage of the process, no expert system shell is suitable and that a system must be developed with programming languages such as LISP, PROLOG, or some conventional programming language.

A multiple discriminant analysis of vessel accidents.

A large sample of 936 vessel accident cases occurring between 1979 and 1987 on the lower Mississippi River were cluster analyzed to generate four groups relatively unique in their respective attribute values. The attributes used to cluster the accidents included participation in the U.S. Coast Guards New Orleans Vessel Traffic Service (NOLA-VTS), type of accident, river stage traffic level, system utilization, accident location, weather conditions, and time of accident. The four-group cluster solution resulted in logical groupings, given the realities of navigating the lower Mississippi River. The four groups resulting from the cluster analysis were characterized as Group 1: Danger Zone (224 cases), 100% NOLA-VTS participants whose accidents occurred primarily on the most dangerous part of the river; Group 2: Bad Conditions for Good Navigators (230 cases), characterized by a high rate of participation and unserious accidents occurring in treacherous navigating conditions; Group 3: Probably Preventable (134 cases), characterized by a low participation rates and serious accidents occurring in not the worst navigating conditions; and Group 4: Accidents That Should Not Have Happened (345 cases), characterized by zero participation and serious accidents occurring in reasonable navigating conditions. Significant marginal participation rates for the marine tracking technology across the four accident clusters (100% for Group 1, 67% for Group 2, 37% for Group 3, and 0% for Group 4) effectively distinguishes between casualty groups. In the subsequent discriminant analysis, three discriminant functions correctly classified 96% of the total accidents, including 100% of Group 1 and Group 4, 90% of Group 2, and 88% of Group 3. The variables contributing most to overall group differentiation were participation in the system, overall system utilization, river stage, traffic level, time and location of accidents. The three discriminant functions were statistically significant, with each individual function accounting for a large relative percentage of the variance between the groups. In order of decreasing discriminating power, the functions could be characterized as (1) System Participation and Utilization, (2) Navigating Conditions, and (3) Time and Place.

A decision support system for academic advising

A decision support system (DSS) was constructed to assist the academic advising staff of a business school. The microcomputerbased system identifies any remaining unsatisfied degree program requirements, selects courses in which the student is eligible to enroll, and prioritizes the courses. The DSS permits advisors to spend time on the more substantive advising issues, such as choice of electives, etc. The system permits students to obtain an optimized listing of courses without the assistance of a human advisor in about five minutes. A high-end spreadsheet (i.e., DSS generator) permits a workable and effective academic advising DSS. The database is the most significant part of this DSS. And, since the modeling component is difficult to separate from the structure of the data itself, a database management system is suggested as a possible better choice as the DSS generator. This platform would provide a more flexible user interface as well as superior data handling capability but at some sacrifice in cost and implementation time.

Pattern development for vessel accidents: a comparison of statistical and neural computing techniques

Abstract This paper describes a sample of over 900 vessel accidents that occurred on the lower Mississippi River. Two different techniques, one statistical and the other based on a neural network model, were used to build logical groups of accidents. The objective in building the groups was to maximize between-group variation and minimize within-group variation. The result was groups whose records were as homogenous as possible. A clustering algorithm (i.e., a non-inferential statistical technique) generated sets of three, four and five groups. A Kohenen neural network model (i.e., a self-organizing map) also generated sets of three, four and five groups. The two sets of parallel groups were radically different as to the relative number of records in each group. In other words, when the two sets of groups were constructed by the respective techniques, the membership of each comparable group within the two different sets was substantially different. Not only was the respective record count in each group substantially different, so were the descriptive statistics describing each comparable set of groups. These results have significant implications for marine policy makers. Important policy variables include safety factors such as weather, speed of current, time of operation, and location of accidents, but mandatory utilization of a voluntary vessel tracking service may be subject to debate.

A structured approach to the evaluation and selection of CASE tools

This paper illustrates an evaluation and selection methodology for CASE software or CASE tools. The methodology’ incorporates three stages 1) CASE software acreenin~ 2) CASE tool evaluation; and, 3) assurance of final CASE software selection. Initially, developing a short list through screening of commercial CASE ptuducta determines whether appropriate tools exist and narrows the field of available CASE software products for detailed conaideralion. The second stage determines which of the remaining products (the finalists) beat meets the needs of the organization, from both functional and technical perspectives. Tle final stage compares user requitvmenta with the features of the selecied CASE software by defining how these requirements will be satisfied by building system application with the selected prrxluct. The methodology also conaidera the pmsaibility that, at any stage of the process, no single CASE product is suitable and that a combination of products must be utilized. Introduction Overview of CASE Computer Aided SYslems Engineering (CASE) has oecupicd a position of prominence and has generated much debate over the past several years. In the midst of all of the “hype” and attention, CASE seems 10 be one of those ill-defined and often misunderstood information system acronyms. Case !echnology, with its wide variety of tools and techniques produced by so many different vendors, seems to have obscured the underlying concepts concomitant with the adoplion of any form of CASE soflware. The primary reason(s) for the adoption of CASE should be built around the need fo~ 1) increased integration of cross-functional syatem~ 2) the improved quality of systems that are developed; and, 3) the integration of business goal

A phased approach to the evaluation and selection of CASE tools

objective% and functions with the systems developed. The adoption of CASE technology not oniy intrcxiuces technological change, but more importantly introduces a change in the basic philosophy of systems development. In some instances the organization wiii experience a “culture shwk’ when embarking upon the implementation of the G4SE environment. The adoption of CASE technology usually demands an indepth evaluation of the organization’s current systems development methodology. Methodologies, such as METHOD/1, NAVIGATOR STWIS, IEF, and IEM, are all characterized by the intnxtuction Permission to copy without fee all or part of this materiel is granted provided that the copiaa ara not made or distributed for diract commercial advantage, the ACM copyright notice and the titla of the publication and ita date appear, and notice ia given that copying ia by permission of the Association for Computing Maohinary. To copy otherwise, or to republish, requirea a fee and/or specific permiaaion. O 1992 ACM 0.89791 -502. x/92/~ 2/1064 ...

An assessment of DSS performance: the impact of utilization and closure

l .50 of “rigor” into ihe systems development prcxeaa. Ail of the aforementioned methodologies have very detailed, specific prwedurcs for systems development. Theac methodologies may be integrated very closely with a specific CASE prcduct, or they may be supported by a variety of products. In either instance, there should be an accepted, specifiq rigorous systems development methodology for the CASE tool to support. AU of the toots and techniques that support the syatemadevelopment methodology (often termed Systems Development Ufe Qcle) and all of ita phases are considered within the realm of CASE software. Moat all methodologies will have the followingmajor phases 1) analysi

An empirical assessment of choice models for software selection: a comparison of the LWA and MAUT techniques

2) design; 3) construction; 4) implementation; and 5) maintenance.