Faculty Research Abstracts
|
|
Ralph D. Badinelli (top)
Title: Multiple-price Yield Management Policies for Inventory Control
Author: Ralph D. Badinelli, Virginia Tech.
Abstract: In support of the increasing interest in integrated planning systems, this paper joins inventory-control with marketing decisions by optimizing the allocation of inventory to different market segments. This allocation can be considered a form of yield management as the market segments are differentiated by the unit selling price and the unit stockout penalty. In effect, the allocation sets "booking limits" for each market segment in each time period over an inventory cycle. In treating inventory allocation in this way we expand the problem of setting safety stock in much the same way that yield management expands the overbooking problem. For the practitioner at the operational level, this research provides practical algorithms for setting the allocation policy. At a strategic level, this research instigates a review of marketing and buffer-stock policies and promotes the use of buffer markets.
Title: Dynamic policies for forecast updating and supply-chain control
Author: Ralph D. Badinelli, Virginia Tech.
Abstract: Forecast updating can introduce complex, dynamic behavior into a forecast independently of such behavior in the actual demand process. This paper examines the interaction of supply-chain policies with the dynamics of forecast updating. Options for configuring supply-chain policies to adapt to forecast updating are investigated.
Dean C. Chatfield (top)
Title: The Bullwhip Effect: Impact of Stochastic Lead Times, Information Quality, and Information Sharing- A Simulation Study.
Authors: Dean C. Chatfield, Virginia Tech; Jeon G. Kim, Penn State; Terry P. Harrison, Penn State; Jack C. Hayya, Penn State.
Abstract: We use a simulation model called ‘SISCO’ to fathom the effects in supply chains of stochastic lead times and of information sharing and level. We test at the outset the accuracy of the simulation by verifying the results in Chen et al. (2000) and Dejonckheere et al. (2003). We find that lead times exacerbate variance amplification in a supply chain and that information sharing and quality are highly significant. For example, using the assumptions in Chen et al. (2000) and Dejounkeree et al. (2002), we find in a numerical experiment of a customer-retailer-wholesaler-distributor-factory supply chain that variance amplification is attenuated by nearly 50 percent at the factory due to information sharing.
Title: Quantifying the Bullwhip Effect in a Supply Chain with Stochastic Lead Times
Authors: Jeon G. Kim, Penn State; Dean C. Chatfield, Virginia Tech; Terry P. Harrison, Penn State; Jack C. Hayya, Penn State.
Abstract: In a recent paper, Dejonckheere, Disney, Lambrecht, and Towill (2003) used control systems engineering (transfer functions, frequency response, spectral analysis) to quantify the bullwhip effect. In the present paper, we, like Chen, Ryan, Drezner, and Simchi-Levi (2000), use the statistical method. But our method differs from Dejonckheere et al. and Chen et al. in that we include stochastic lead time and provide exact results in quantifying the bullwhip effect, both with information sharing and without information sharing. We use iid demands in a k-stage supply chain for both. By contrast, Chen et al. provide lower bounds using autoregressive demand for information sharing and for information not sharing (with zero safety factor for stocks). Dejonckheere et al. validate Chen et al.’s results for a 2-stage supply chain without information sharing, using both autoregressive and iid normally distributed demands. More significantly, we differ in the method of estimating mean demand during lead time, with our method, based on interviews with supply chain managers, yielding lower variance amplification than Chen et al., whose estimation method relied on mangers of a large retail company. As for the effect of information, we find that the variance increases nearly linearly in echelon stage with information sharing and exponentially in echelon stage without information sharing. We verify our analytical results by simulation.
Title: Supply Chain Information Technology Metrics
Authors: Jack C. Hayya, Penn State; Chao-Hsien Chu, Penn State; Dean C. Chatfield, Virginia Tech; Xin (James) He, Fairfield University.
Abstract: In designing integrated supply chain information architecture, it is necessary to provide metrics for the different components of the system. Here, we present the metrics for security, for graphical user interface, for the hardware and the software, and for information and storage. Also relevant are metrics for the validation of the information technology (IT) architecture. The goal of this research is to provide a framework for measuring the performance of supply chain IT and to motivate more work on IT metrics in order to fill some gaps on this subject in the literature. The analysis is supported by work on this topic with the military and by interviews with executives from a Korean steel manufacturer and a large U.S. defense firm.
Title: Order Flow in Serial Supply Chains
Authors: Dean C. Chatfield, Virginia Tech; Jeon G. Kim, Penn State; Terry P. Harrison, Penn State; Jack C. Hayya, Penn State.
Abstract: We capture order flows in supply chains and fathom the makings of the bullwhip effect. We investigate a serial supply chain, using a new simulation engine (SISCO), developed by Chatfield (2001). We begin with a serial supply chain with five nodes -- customer demand, retailer, wholesaler, distributor, factory -- and then verify the simulation results mathematically. The customer demands are normally distributed, but those at the other nodes are not be necessarily so because of the convolution with lead time. We find that if the order system parameters remain unchanged there would be no Bullwhip Effect. But with human intervention in continual updating of system parameters, there would be an amplification of variance as we go up the supply chain. We also find that the use of the normal approximation to lead-time demand in setting up safety stocks at the upper nodes can be egregiously wrong. The Bullwhip Effect is exponential with echelon if customer demand information is not shared; it is quadratic when customer demand information is shared.
Title: The Economic Lot Scheduling Problem: A Hybrid Genetic Search Approach
Authors: Dean C. Chatfield, Virginia Tech.
Abstract: The economic lot scheduling problem (ELSP) is the challenge of accommodating several products to be produced on a single machine in a cyclical pattern. A solution involves determining a repetitive production schedule for n products with a goal of minimizing the total of setup and holding costs. We develop a procedure that combines a new solution structure, a hybridized item scheduling approach, and a genetic algorithm search procedure for determining a production schedule. This procedure is the first to explicitly state the assignment of products to periods as part of the solution structure. The procedure was applied to a six-problem test suite, including Bomberger's (1966) stamping problem, a problem that has been under investigation for over 35 years. The genetic lot scheduling procedure produced impressive results, including the best solutions obtained to date on some problems.
Deborah F. Cook (top)
Title: Using Radial Basis Function Neural Networks to Recognize Shifts in Correlated Manufacturing Process Parameters.
Authors: Cook, D. F., Virginia Tech, and C. C. Chiu, Fu-Jen Catholic University.
Journal: IIE Transactions 30(3) 227-234, 1998.
Abstract:Traditional statistical process control (SPC) techniques of control charting are not applicable in many process industries because data from these facilities are autocorrelated. Therefore the reduction in process variability obtained through the use of SPC techniques has not been realized in process industries. Techniques are needed to serve the same function as SPC control charts, that is to identify process shifts, in correlated parameters. Radial basis function neural networks were developed to identify shifts in process parameter values from papermaking and viscosity data sets available in the literature. Time series residual control charts were also developed for the data sets. Networks were successful at separating data that were shifted 1.5 and 2 standard deviations from nonshifted data for both the papermaking and viscosity parameter values. The network developed based on the papermaking data set was also able to separate shifts of 1 standard deviation from nonshifted data. The SPC control charts were not able to identify the same process shifts. The radial basis function neural networks can be used to identify shifts in process parameters, thus allowing improved process control in manufacturing processes that generate correlated process data.
Title: Combining a Neural Network with a Genetic Algorithm for Process Parameter Optimization.
Authors: Cook, D. F., Virginia Tech, C. T. Ragsdale, Virginia Tech, and R. L. Major, Virginia Tech.
Journal: Engineering Applications of Artificial Intelligence 13(4) 391-396. 2000.
Abstract:A neural-network model has been developed to predict the value of a critical strength parameter (internal bond) in a particleboard manufacturing process, based on process operating parameters and conditions. A genetic algorithm was then applied to the trained neural network model to determine the process parameter values that would result in desired levels of the strength parameter for given operating conditions. The integrated NN-GA system was successful in determining the process parameter values needed under different conditions, and at various stages in the process, to provide the desired level of internal bond. The NN-GA tool allows a manufacturer to quickly determine the values of critical process parameters needed to achieve acceptable levels of board strength, based on current operating conditions and the stage of manufacturing.
Title: An Augmented Neural Network Classification Approach to Detecting Mean Shifts in Correlated Manufacturing Process Parameters.
Authors: Zobel, C. W., Virginia Tech, Cook, D. F., Virginia Tech, and Q. J. Nottingham, Virginia Tech.
Journal: International Journal of Production Research. In Press.
Abstract: Statistical process control (SPC) techniques have traditionally been used to identify when the mean of a manufacturing process has shifted out of control. In situations where there is correlation among the observed outputs of the process, however, the underlying assumptions of SPC are violated and alternative approaches such as neural networks become necessary in order to characterize the process behavior. This paper discusses combining a neural network based procedure with a graphical classification technique to identify shifts in a process mean. This approach improves upon the performance of previous neural network and statistical techniques.
Kellie Keeling (top)
Title: A Regression Equation for Determining the Dimensionality of Data
Author: Kellie B. Keeling, Virginia Tech.
Journal: Multivariate Behavioral Research, Vol. 35, No. 4, 2000 p. 457-468.
Abstract: Parallel analysis has received much support and attention as a criterion for using eigenvalues to determine the dimensionality of data. Parallel analysis compares sample eigenvalues to expected eigenvalues of a sample from a correlation matrix generated by independent normally distributed random variables. To make parallel analysis more accessible to researchers, several studies have proposed multiple regression equations for estimating the expected value of the eigenvalues of a sample correlation matrix assuming that the population correlation matrix is the identity matrix. A new regression equation to estimate the mean value of eigenvalues is presented in this paper and a comparative study reveals favorable performance of this proposed equation to previously published regression equations. This proposed technique has the advantage that a table of coefficients, listing regression coefficients for each eigenvalue root, is not needed.
Title: Assessing the Impact of Different Versions of Nonparametric Procedures and a Proposed Simplified Implementation using Excel
Authors: Robert J. Pavur, University of North Texas, Kellie B. Keeling, Virginia Tech.
Journal: Proceedings of the Southeastern Decision Sciences Institute, February 2000.
Abstract: Over the past decade, many introductory statistics books have been written with Microsoft Excel as the primary computer package for analyzing data. This shift in emphasis to Excel as the primary computer package that business students may be using has prompted professors to become innovative in making Excel do more than the standard features it offers. One statistical area that is clearly absent from Excel's Data Analysis Tools is nonparametric statistics. This paper illustrates how nonparametric procedures can be easily implemented by inputting ranks of the data into Excel's parametric procedures. A complication in using nonparametric procedures is that their observed significance levels are approximated and may or may not be corrected for ties. Another complication is that Excel's ranking function does not have an option of assigning average ranks. An example using the Kruskal Wallis test demonstrates how various approaches can yield different conclusions. A simulation study with the Kruskal Wallis statistic and various approximations illustrates the sensitivity of the test procedures to ties in the ranks.
Title: Examining E-Commerce Oriented Web Home Pages: A Comparative Study Among Fortune 500 Companies
Authors: Xiaoni Zhang, University of North Texas, Kellie B. Keeling, Virginia Tech, and Robert J. Pavur, University of North Texas.
Journal: Information Systems Research, Submitted January 2001.
Abstract: Fortune 500 companies have often provided leadership in marketing products. E-commerce is now forcing companies to think of how their company's image, products, and leadership are affected by e-commerce Web sites. This study searches for substantive relationships between information quality and characteristics of Web home pages among companies that have an e-commerce orientation and those that do not. The success of e-commerce may depend on user perceptions of a Web site. The home page typically gives the first impression. This research explores user perceptions of presentation, navigation, and quality of Web home pages for approximately 200 selected Fortune 500 companies across 10 industries. An instrument is developed to measure these constructs and is assessed for convergent and discriminant validity as well as reliability. Company Web home pages are clustered using 24 Web site features. Four clusters were identified with only one cluster being clearly e-commerce oriented. This cluster is shown to be rated the highest in user perceptions of presentation, navigation, and quality of Web home pages. An analysis of the data provides valuable insight into trends in e-commerce among various industries that make up the Fortune 500.
Patty L. Kitchin (top)
Title: A New Method For Comparing Experiments And Measuring Information
Authors: Patty L. Kitchin, Virginia Tech, and R. V. Foutz, Virginia Tech (Statistics).
Journal: Communications in Statistics, Computation and Simulation, 30(1), 2001.
Abstract: A statistical experiment can consist of taking a sample. Often, statistics are formed that are based on the sample data. Two different experiments can yield two different statistics. Under certain conditions, we can compare these statistics by comparing the experiments. A statistic that summarizes an entire data set without losing any information about the family of distributions or the model is a sufficient statistic. Sometimes it may be desirable to use a statistic that, though not sufficient, does summarize the data set somewhat. How much information will we lose? How can we compare two statistics that are not sufficient in terms of the amount of information they provide?
A new method for comparing experiments and measuring information is introduced. The new method is used to evaluate the expected efficiency of a statistic in discriminating between any two values of the parameter as compared to a sufficient statistic. This new method can be self-calibrated to give this expected efficiency a meaningful scale. This new method is applied to Casino Blackjack. Several card-counting statistics are compared by the amount of information each provides in discriminating between different deck compositions as compared to a sufficient statistic. This new method provides new insight about information in card-counting statistics by putting this information on a meaningful scale.
Title: Measuring the Amount of Statistical Information in the EPT Index
Authors: Patty L. Kitchin, Virginia Tech
Journal: Environmetrics, in press
Abstract: Biological monitoring is the process of measuring the effect of environmental stress on the environment. Aquatic macroinvertebrates are widely used in the monitoring of freshwater lotic systems. The macroinvertebrate fauna of a reference stream is commonly compared to the fauna of an impacted stream that is affected by an environmental stressor. The smaller the similarity between these two streams, the greater the effect of pollution or stress on the impacted stream. Many richness measures, or statistics, exist for measuring similarity. These statistics can be computed using different levels of taxonomic resolution (species, genus and family). Many aquatic biologists believe that species-level identifications, which require exorbitant time and expertise, are needed for correct data interpretations. The actual amount of information provided by these statistics at different taxonomic levels has never been measured. In this paper the amount of statistical information contained in one particular richness measure, the EPT index, is measured by computing the expected efficiency of this statistic at distinguishing between a reference stream and an impacted stream as compared to a sufficient statistic. This expected efficiency is computed at the various levels of taxonomic resolution.
Title: A derivation of the amount of sufficiency in a statistic.
Authors: R. V. Foutz, Virginia Tech (Statistics), Patty L. Kitchin, Virginia Tech
Journal: to be resubmitted to Journal of Statistical Planning and Inference
Abstract: A basic result of Lehmann and Scheffe provides a method for constructing a minimal sufficient statistic. This article shows that the result can also be used to quantify the amount of sufficiency in a statistic. In particular, the amount of sufficiency that is possessed by X in addition to that possessed by a statistic T (X) is derived from five reasonable axioms. On axiom is motivated by the result of Lehmann and Scheffe. The other axioms resemble those used by Good for deriving an explicatum for information. It is concluded that the amount of sufficiency that is possessed by a statistic T (X) is unique up to a continuous monotonic function. The conclusion also leads to a metric d(T1,T2) for quantifying the exclusiveness between the sufficiency content within T1 (X) and the sufficiency content within T2 (X). The metric is zero if and only if the statistics T1 (X) and T2 (X) possess the identical sufficiency content.
Title: An automated statistical quality control process for Internet service providers
Authors: Patty L. Kitchin, Virginia Tech, D. Novak, University of Connecticut, Clark Gaylord, Virginia Tech (CNS)
Journal: to be submitted to European Journal of Operations Research
Abstract: While the use of statistical quality control (SQC) techniques are very common in manufacturing, they do not appear to be widely used for evaluating and improving the performance of networking hardware for dialup Internet service. This paper discusses the development and implementation of an automated SQC process for a large university dialup Internet Service Provider (ISP). In this paper critical network performance variables are identified, underlying distributional assumptions and the variable transformation process is described, and the statistical procedures used to establish performance baselines and identify out of control processes are presented. This SQC program is then implemented at and evaluated by the communications network services of Virginia Tech.
Lance A. Matheson (top)
Title: Information Technology: A Study of Accountants’ Skills and Knowledge Levels
Authors: McCraw, J. H., University of West Georgia, J. R. O'Malley, Jr., UNC Charlotte, B. M. Bird, University of West Georgia, L. A. Matheson, Virginia Tech
Journal: The Review of Business Information Systems, Vol. 7, No. 3, (2003), pp. 45-48
Abstract: This article examines the results of a survey mailed to 940 accounting professionals in which they identify the categories of IT skills and knowledge that are required to perform their job. Using cluster analysis, this article next examines whether - and to what extent - relationships exist between different categories ot IT skills and knowledge. Survey results are then presented regarding methods by which accounting professionals acquire needed job-related skills and knowledge.
Title: New Directions for Research in Electronic Data Interchange (EDI)
Authors: O'Malley, Jr., J. R., UNC Charlotte, L. A. Matheson, Virginia Tech
Journal: Journal of Information Technology Management, (2006)
Abstract: Electronic Data Interchange was envisioned as the future for business with its ability to lower costs, improve productivity, and decrease transaction times. The adoption of EDI has been very slow compared to other information technologies. This article develops a framework to explain why this has occurred. Uncertain economic benefits and the costs of implementing ever-changing EDI standards are shown to be major reasons. In addition, it also proposes directions for future trends and research in EDI.
Quinton J. Nottingham (top)
Title: Local logistic regression: An application to army penetration data
Authors: Nottingham, Q. J., Virginia Tech, Birch, J. B., Virginia Tech (Statistics) and Bodt, B. A., U.S. Army Research Laboratory.
Journal: Journal of Statistical Computation and Simulation, Volume 66, 35-50
Abstract: There have been a number of procedures used to analyze non-monotonic binary data to predict the probability of response. Some classical procedures are the Up and Down strategy, the Robbins-Monro procedure, and other sequential optimization designs. Recently, nonparametric procedures such as kernel regression and local linear regression have been applied to this type of data. It is a well known fact that kernel regression has problems fitting the data near the boundaries and a drawback with local linear regression is that it may be "too linear" when fitting data from a curvilinear function. The procedure introduced in this paper is called local logistic regression, which fits a logistic regression function at each of the data points. An example is given using United States Army projectile data that supports the use of local logistic regression when analyzing non-monotonic binary data for certain response curves. Properties of local logistic regression will be presented along with simulation results that indicate some of the strengths of the procedure.
Title: Visualization of Multivariate Data with Radial Plots Using SAS
Authors: Nottingham, Q. J., Virginia Tech, Cook, D. F., Virginia Tech, and Zobel, C. W., Virginia Tech.
Journal: accepted for publication to Computers and Industrial Engineering
Abstract: Data visualization tools can provide very powerful information and insight when performing data analysis. In many situations, a set of data can be adequately analyzed through data visualization methods alone. In other situations, data visualization can be used for preliminary data analysis. In this paper, radial plots are developed as a SAS-based data visualization tool that can improve one's ability to monitor, analyze, and control a process. Using the program developed in this research, we present two examples of data analysis using radial plots; the first example is based on data from a particle board manufacturing process and the second example is a business process for monitoring the time-varying level of stock return data.
Title: Local Linear Regression for Estimating Time Series Data
Authors: Nottingham, Q. J., Virginia Tech, and Cook, D. F., Virginia Tech.
Journal: accepted for publication to Computational Statistics and Data Analysis
Abstract:Predicting future performance based on past performance history is a task often undertaken by business process managers. Various statistical and analytical techniques, such as time series and neural network modeling, are available. However, these techniques require the availability of a long time series for the development of a predictive model. Local linear regression (LLR) is an additional nonparametric statistical method that can be used to estimate a time series response variable. The LLR technique does not require a long time series for the development of a predictive model. In fact, the LLR technique can be utilized for prediction once three data points have been collected from the business process. In this work, LLR was evaluated as a tool for predicting future values of process parameters based on historical values. If successful, the LLR technique could be applied in start-up conditions or used as an alternative in some situations to time series modeling. The LLR procedure outperformed traditional time series techniques for the example stationary data sets and had comparable results to the ARIMA model for the example seasonal data set. In addition the LLR technique uses the data that is currently available from a process as its basis for prediction, thus providing a dynamic predictive technique that can continue to function in the presence of process changes.
Cliff T. Ragsdale (top)
Title: The Ordered Cutting Stock Problem
Authors: Ragsdale, C.T., Virginia Tech and C. Zobel, Virginia Tech.
Journal: Decision Sciences, forthcoming.
Abstract: The one-dimensional Cutting Stock Problem (CSP) is a classic combinatorial optimization problem. This paper identifies and discusses a new type of one-dimensional CSP, called the ordered CSP, which explicitly restricts the number of jobs that can be open at any given point in time to a single job. Given the growing emphasis on mass customization in the manufacturing industry, this restriction can help lead to a reduction both in inventory levels and in material handling activities. A formal mathematical formulation is provided for the new CSP model, and its applicability is discussed with respect to a production problem in the custom door and window manufacturing industry. Based upon the structure of the ordered CSP, a genetic algorithm (GA) solution approach is then presented which incorporates a customized repair heuristic for reducing scrap levels. Several different production scenarios are considered, and computational results are provided which illustrate the ability of the GA-based approach to significantly decrease both the amount of scrap generated and the number of pieces of raw stock that are used in the production process.
Title: A Decision Support Methodology for Stochastic Multi-Criteria Linear Programming Using Spreadsheets.
Authors: Novak, D., University of Connecticut, and C. T. Ragsdale, Virginia Tech.
Journal: Decision Support Systems, Vol. 36, No. 1, pp. 99-116, 2003.
Abstract: In recent years, tools for solving optimization problems have become widely available through the integration of optimization software (or solvers) with all major spreadsheet packages. These solvers are highly effective on traditional linear programming (LP) problems with known, deterministic parameters. However, thoughtful analysts may rightly question the quality and robustness of optimal solutions to problems where point estimates are substituted for model parameters that are stochastic in nature. Additionally, while many LP problems implicitly involve multiple objectives, current spreadsheet solvers provide no convenient facility for dealing with more than one objective. This paper introduces a decision support methodology for identifying robust solutions to LP problems involving stochastic parameters and multiple criteria using spreadsheets.
Title: A Decision Support System for the Electrical Power Districting Problem.
Authors: Bergey, P., North Carolina State University, C. T. Ragsdale, Virginia Tech, Mangesh Hoskote, The World Bank.
Journal: Decision Support Systems, Vol. 36, No. 1, pp. 1-17, 2003.
Abstract: Many national electricity industries around the globe are being restructured from regulated monopolies to deregulated marketplaces with competitive business units. The business units responsible for transmission and distribution must be given physical property rights to certain parts of the power grid in order to provide reliable service and make effective business decisions. However, partitioning a physical power grid into economically viable districts (distribution companies) involves many considerations. We refer to this complex problem as the electrical power districting problem (EPDP). This research identifies the fundamental characteristics required to appropriately model and solve an EPDP. The proposed solution methodology is implemented as a decision support system (DSS) featuring a visualization tool that allows decision makers (DMs) to explore what we refer to as a "soft efficient frontier." This DSS was found to effectively support DMs at The World Bank in solving an EPDP in the context of a case study for the Republic of Ghana.
Title: Scheduling Pre-Printed Newspaper Advertising Inserts Using Genetic Algorithms.
Authors: Carter, A., Radford University and C. T. Ragsdale, Virginia Tech.
Journal: OMEGA: The International Journal of Management Science, Vol. 30, No. 6, pp. 415-421, 2002.
Abstract: In recent years, the use of pre-printed advertising inserts in newspapers has increased dramatically. Pre-printed inserts allow advertisers to deliver colorful, high-quality marketing material to targeted groups of consumers within the newspaper's delivery zone structure. To accommodate the increased workload associated with pre-printed inserts without negatively impacting the news deadline or delivery schedules, many newspaper companies face increasingly complex post-press scheduling decisions. This paper presents a spreadsheet model developed to represent the pre-printed insert scheduling problem in a case study of an actual medium-size newspaper company. The performance of two commercial genetic algorithm (GA) optimizers is compared on this problem. Computational testing shows the GAs develop schedules that substantially reduce the post-press production department's insert processing time.
Title: The Competitive Market Efficiency of Hotel Brands: An Application of Data Envelopment Analysis.
Authors: Brown, J., Virginia Tech, and C. T. Ragsdale, Virginia Tech.
Journal: Journal of Hospitality & Tourism Research, Vol. 26, No. 4, pp. 332-360, 2002.
Abstract: The objective of this paper is to illustrate how managers in the hotel industry might analyze and improve their brands' market efficiency using data envelopment analysis (DEA). We evaluated competitive market efficiency in terms of customer satisfaction and customer value. Our DEA results show that 23 of the 46 hotel brands studied were efficient. The inefficient hotels generated less customer satisfaction and customer value for the same level of inputs relative to their more efficient competitors. In particular, the competitive market inefficient hotels suffered more guest complaints; employed a less helpful and efficient check-in staff; did not maintain their rooms, grounds, and public spaces acceptably; and charged higher prices than justified by their market offerings. Further, hotel chains that were not competitive market efficient had a less than optimal number of properties in the chain as well as a less than optimal number of hotel rooms. Our results illustrate how managers can use DEA to improve the relative market efficiency of their brands.
Christopher W. Zobel (top)
Title: An augmented neural network classification approach to detecting mean shifts in correlated manufacturing process parameters
Authors: C.W. Zobel, D.F. Cook, and Q.J. Nottingham, Virginia Tech
Journal: Accepted for publication in the International Journal of Production Research, 2003.
Abstract: Statistical process control (SPC) techniques have traditionally been used to identify when the mean of a manufacturing process has shifted out of control. In situations where there is correlation among the observed outputs of the process, however, the underlying assumptions of SPC are violated and alternative approaches such as neural networks become necessary in order to characterize the process behavior. This paper discusses the development of a neural network technique that provides a significantly improved capability for recognizing these process shifts as compared to the current techniques in the literature. The procedure in question is an augmented neural network based approach which incorporates a data preprocessing classification algorithm that provides information to facilitate early detection of out of control operating conditions. This approach is shown to improve significantly upon the performance of previous neural network techniques for identifying process shifts in the presence of correlation.
Title: Utilization of Neural Networks for the Recognition of Variance Shifts in Manufacturing Process Parameters
Authors: D.F. Cook, C.W. Zobel, and Q.J. Nottingham, Virginia Tech
Journal: International Journal of Production Research, vol. 39, no. 17 (2001), pp. 3881-3887.
Abstract: Traditional statistical process control charting techniques were developed for use in discrete industries where independence exists between process parameter values over time. Process parameters from many manufacturing industries are not independent, however; they are serially correlated. Consequently, the power of traditional statistical process control charts is greatly weakened. This paper discusses the development of neural network models to successfully identify shifts in the variance of correlated process parameters. These neural network models can be used to monitor manufacturing process parameters and to signal when process adjustments are needed.
Title: Simulation-based Policy Generation using Large-scale Markov Decision Processes
Authors: C.W. Zobel , Virginia Tech, and W.T. Scherer, University of Virginia.
Journal: IEEE Transactions on Systems, Man & Cybernetics - Part A: Systems and Humans, vol. 31, no. 6, Nov. 2001, pp. 609-622.
Abstract: This paper presents a new problem-solving approach, termed Simulation-based Policy Generation (SPG), that is able to generate solutions to problems that may otherwise be computationally intractable. When trying to optimize large-scale sequential stochastic problems, it is often easier to simulate the system under consideration and then to perform some type of simulation/optimization procedure, such as pseudo-random search or a response surface methodology. The SPG method builds on this idea by using a simulation of the original problem to create an approximating Markov decision process (MDP) model which is then solved via traditional MDP solution approaches. Since this approximating MDP is a fairly rich and robust sequential optimization model, solution policies can be created which represent an intelligent and structured search of the policy space. An important feature of the SPG approach is its adaptive nature, in that it uses the original simulation model to generate improved aggregation schemes, allowing the approach to be applied in situations where the underlying problem structure is largely unknown. In order to illustrate the performance of the SPG methodology, we apply it to a common but computationally complex problem of inventory control, and we briefly discuss its application to a large-scale telephone network routing problem.