L. Douglas Smith

Prototyping for systems development: a critical appraisal

Although relatively recent in information systems design, the prototyping technique has a long tradition in developing engineering systems. An engineering system is defined as any artificial system that performs actions to achieve a desired transformation of objects undergoing a change of state. We will review the application of and experience with prototyping in engineering systems design and relate these to the development of information systems. Drawing on the strong similarities between the design processes of engineering systems and information systems results in the identification of a number of different types of prototypes that can be used for a variety of purposes and integrated into the various stages of the systems development life cycle. The experience gained from applying the prototyping method in the design of engineering systems permits us to exploit its advantages and to avoid its misapplications when it is used in information systems development.

A computer-based nurse scheduling system

Abstract This paper describes a computer-based system for scheduling nurses in a St. Louis hospital. List processing and problem oriented data structures provided a flexible framework for the development of a heuristic which considers a complicated set of constraints when generating monthly shift schedules. The scheduling system covers several staff categories (supervisors, RNs, LPNs, AIDES, secretaries), considers individual preferences for shifts and days off, includes part-time employees, accommodates special requests for days off or particular shift assignments, and provides a convenient interface for the scheduling clerks who make final adjustments to the computer-generated schedules.

An analysis of default risk in mobile home credit

Abstract There are approximately US

Forecasting losses on a liquidating long-term loan portfolio

22 billion in outstanding receivables and an active secondary market for mobile home credit in the United States. This important segment of the financing industry shares some attributes with traditional home mortgage financing and other attributes with consumer credit. In this paper we draw on the mortgage literature to postulate the determinants of default risk for mobile home loans. Using multivariate logit models, we investigate the relevance of payment history, loan terms, borrower characteristics, economic conditions, and legal constraints in analyzing loan defaults and delinquencies. Payment history emerges as the overwhelming factor in predicting the likelihood of default.

Strategic planning for data warehousing

Abstract Assessing the condition of financial institutions and valuation of loan portfolios in secondary markets require the estimation of exposure to losses on existing portfolios of long-term loans. Data from a major U.S. financial institution were used to construct a forecasting model with Markovian structure and nonstationary transition probabilities. The model proved to be effective in representing changes in probability of default that occur as individual loans mature and accurate in forecasting aggregate defaults and losses on a nationwide portfolio of long-term loans.

Linear Transformations of Polynomial Regression Models

Abstract In the course of a major strategic planning exercise for a large municipal government, data warehousing emerged as a tactic for consolidating and sharing information among many City departments with different informational needs and a variety of computing platforms. In this paper, we describe processes and analytical methods for shaping a data warehousing strategy and for determining the contents of the data warehouse. We also present some cautionary conclusions about the development of monolithic data warehouses to serve an entire organization.

Allocating inter-station inventory capacity in unpaced production lines with heteroscedastic processing times

Abstract Statistical texts differ in the ways they test the significance of coefficients of lower-order terms in polynomial regression models. One reason for this difference is probably the concern of some authors that t ratios for the regression coefficients of lower-order terms may change as a result of linear scale transformations of the independent variables. Empiricists generally report tests of these terms and attempt to interpret their meanings. This paper succinctly demonstrates how linear transformations of polynomial regression models affect tests of statistical significance for individual parameters.

Assessment of business potential at retail sites: empirical findings from a US supermarket chain

This paper demonstrates problems in allocating inter-station inventory capacities in unpaced production lines with heteroscedastic processing times. Simulations were performed to show the effects of total work-in-process inventory capacity, relative inter-station inventory allocations and pattern of processing time variances upon production rate, work station idle times and time to complete a product. Trade offs among the three performance measures were observed and deficiencies in intuitive strategies for redesigning lines were deduced.

A set-processing algorithm for scheduling staff on 4-day or 3-day work weeks

We address the problem of estimating retail business potential at alternative sites, with concern for assessing performance relative to potential in existing markets and for identifying the best sites for expansion into new markets. At question is the utility of information typically used in formal retail patronage models, in comparison with additional information considered important by retail executives. Relevant data are gathered from secondary sources and intensive in-store surveys are conducted to produce a portfolio of information about neighbourhood demographics, store ambience, variety and quality of products and services, relative prices of selected products, etc. for stores in a retail grocery chain and competitive stores in the chains markets. We experiment with alternative statistical models for store performance to determine the consequence of restricting the types of data available when constructing the models. Our findings suggest that while information about store location and surrounding areas, store characteristics and competitive position are all required to obtain the best assessment for business potential at a site, a few key variables on each dimension offer the bulk of explanatory power. Further, the spatial-locational variables affect all measures of store performance in intuitive directions, whereas the effects of other variables differ according to performance measure and reflect the stores market position.

Strategic Planning for Municipal Information Systems: Some Lessons from a Large U.S. City

This paper presents an efficient algorithm for scheduling a single-category work force on 4-day or 3-day work weeks. Employees work 4 or 3 days each week, have A out of every B weekends off, and work no more than 5 consecutive days in a work stretch on 4-day work weeks and no more than 4 days in a work stretch on 3-day work weeks. Such conditions often prevail in 7-day-a-week organizations such as hospitals, manufacturing plants, and retail stores. We determine the minimum number of workers required to satisfy the scheduling constraints under any pattern of daily requirements. Then we present the algorithm for assigning days off for each worker, thereby determining the work schedules. We show that the algorithm, by construction, will necessarily satisfy the scheduling constraints.