Two-Group Classification Using Neural Networks*

Artificial neural networks are new methods for classification. We investigate two important issues in building neural network models; network architecture and size of training samples. Experiments were designed and carried out on two-group classification problems to find answers to these model building questions. The first experiment deals with selection of architecture and sample size for different classification problems. Results show that choice of architecture and choice of sample size depend on the objective: to maximize the classification rate of training samples, or to maximize the generalizability of neural networks. The second experiment compares neural network models with classical models such as linear discriminant analysis and quadratic discriminant analysis, and nonparametric methods such as k-nearest-neighbor and linear programming. Results show that neural networks are comparable to, if not better than, these other methods in terms of classification rates in the training samples but not in the test samples.     

ON THE IMPLICATIONS OF SPECIFICATION UNCERTAINTY IN FORECASTING*

Forecasters typically select a statistical forecasting model from among a set of alternative models. Subsequently, forecasts are generated with the chosen model and reported to management (forecast consumers) as if specification uncertainty did not exist (i.e., as if the chosen model were the “true” model of the forecast variable). In this note, a well-known Bayesian model-comparison procedure is used to illustrate some of the ambiguities and distortions of forecasts that do not reflect specification uncertainty. It is shown that a single selected forecasting model (however chosen) will generally misstate measures of forecast risk and lead to point and interval forecasts that are misplaced from a decision-theoretic point of view.     

An Inventory Model with Planned Shortages and Price Dependent Demand

Previous research has yielded a procedure for a retailer to determine the optimal lot size and selling price when a supplier offers all-unit quantity discounts and demand is a decreasing function of price. In this paper, we extend that research by allowing for shortages. An algorithm is presented that determines the optimal lot size, order level, and selling price for a class of demand functions, including the constant price-elasticity and linear demand functions.     

A Comparison of Reorder Point and Material Requirements Planning Inventory Control Logic*

This paper reports the results of simulation experiments that compared the inventory efficiency (i.e., the customer service level provided by a given level of inventory) for two different inventory policies. One of these policies uses time-phased information on future demands like that found in material requirements planning (MRP) systems, while the other (the reorder point or ROP policy) relies on forecasts implicitly based on average past demands. After establishing that the MRP policies dominate for reasonable conditions, the uncertainty of the forecasts was manipulated until the policy preference was reversed. It requires a very perverse relationship between the forecast and actual demand before ROP beats MRP on inventory efficiency.     

On the Repeated Test-Retest Reliability of the End-User Computing Satisfaction Instrument: A Comment*

This comment extends the test-retest reliability of the end-user computing satisfaction (EUCS) instrument by Torkzadeh and Doll [10]. Whereas Torkzadeh and Doll [10] reportedstability for two hour and two week EUCS test-retest reliability, we investigate the test-retest reliability of the EUCS instrument at two points in time, separated by a two yearinterval. We assess the end user computing satisfaction of personal computer, as well as mainframe, administrative end users in a large public organization. The results of the repeated test-retest using differing application platforms add further support for the reliability of the EUCS measure and highlight some areas of concern for managers of information systems.     

Multiechelon (R,S) Inventory Model*

Most studies in multiechelon inventory systems have concentrated on understanding the specific aspects of a system's behavior. The problem of optimal policy computation has largely been ignored. In this paper, we investigate a two-echelon inventory system experiencing stochastic demand and a pull system of inventory allocation. Both echelons use an order-up-to-level type control policy. A mathematical model is developed to determine the optimal order level at all echelons and validated through simulation. Two simple algorithms to locate the optimum solution are presented. The use of graphical tools in optimal policy calculation is also discussed.     

Inventory Control and Scheduling of Multiple Products in a Common Facility*

Consider a set of chemical products to be produced in a single facility. Each product has its own unique reaction time (which is assumed to be independent of its batch size), as well as other cost and demand values. In this paper, we address the problem of determining the optimal number of batches, batch sizes, and an accompanying production schedule for these products in the single facility that will minimize the total cost. Two different algorithms have been developed for this problem, the performances of which are contrasted with classical cyclic production schedules. Finally, some guidelines for the application of these methods to real-life problems are outlined.     

Improving Revenue Management Decision Making for Airlines by Evaluating Analyst‐Adjusted Passenger Demand Forecasts*

To maximize revenue, airline revenue management analysts (RMAs) attempt to protect the right number of seats for late‐booking, high‐revenue‐generating passengers from low‐valued leisure passengers. Simulation results in the past showed that a major airline can generate approximately $500 million per year through efficient RM operations. Accurate passenger demand forecasts are required, because reduced forecast error significantly improves revenue. RMAs often adjust the system forecasts to improve revenue opportunity. Analysis of system forecast performance and analyst adjustment is complex, because one must account for all unseen demands throughout the life of a flight. This article proposes a method to account for unseen demand and evaluate forecast performance (adjusted or unadjusted) through a forecast monitoring system. Initial results from one major airline's origin‐destination market data justify the value of RMA forecasting adjustments.     

Supporting Quick Response Through Scheduling of Make-to-Stock Production/Inventory Systems*

Scheduling of traditional job shops in make-to-order systems has seen extensive research over the past three decades. In such systems, performance is often related to various job completion metrics such as average flow time, average lateness, etc. This paper examines a scheduling problem in a make-to-stock environment where individual job completion measures are irrelevant. In this case, customer orders are satisfied through on-hand inventory where customer service is more closely related to the manufacturer's ability to quickly satisfy demand. We consider the role of scheduling in reducing inventories and improving customer service in the context of a manufacturer who assembles several different products on a single assembly line. We develop scheduling rules for such a system and experimentally compare their performance to those typically used in such environments. Our results indicate that rules which consider the inventory position and demand forecast outperform traditional fixed cycle rules.     

Economic Production Lot Sizing with Periodic Costs and Overtime*

Traditional approaches for modeling economic production lot‐sizing problems assume that a single, fixed equipment setup cost is incurred each time a product is run, regardless of the quantity manufactured. This permits multiple days of production from one production setup. In this paper, we extend the model to consider additional fixed charges, such as cleanup or inspection costs, that are associated with each time period's production. This manufacturing cost structure is common in the food, chemical, and pharmaceutical industries, where process equipment must be sanitized between item changeovers and at the end of each day's production. We propose two mathematical problem formulations and optimization algorithms. The models' unique features include regular time production constraints, a fixed charge for each time period's production, and the availability of overtime production capacity. Experimental results indicate the conditions under which our algorithms' performance is superior to traditional approaches. We also test the procedures on a set of lot‐sizing problems facing a national food processor and document their potential economic benefit.     

On Inventory Abatement via Manufacturing Randomness Reductions*

Under a continuous improvement framework, the policy of abating inventory via reductions in manufacturing randomness is considered. To explore this policy, a model of a real-world production-inventory system is developed, tested, and studied. The results suggest that manufacturing randomness reductions, even substantial ones, may not necessarily lead to inventory abatement, and paradoxically may lead sometimes to an inventory increase. In these cases, however, manufacturing randomness reductions will translate into higher customer service levels.     

A Decision Model for the Robot Selection Problem Using Robust Regression*

Industrial robots are increasingly used by many manufacturing firms. The number of robot manufacturers has also increased with many of these firms now offering a wide range of models. A potential user is thus faced with many options in both performance and cost. This paper proposes a decision model for the robot selection problem. The proposed model uses robust regression to identify, based on manufacturers' specifications, the robots that are the better performers for a given cost. Robust regression is used because it identifies and is resistant to the effects of outlying observations, key components in the proposed model. The robots selected by the model become candidates for testing to verify manufacturers' specifications. The model is tested on a real data set and an example is presented.     

EMPIRICAL INVENTORY SIMULATION: A CASE STUDY*

An empirical simulation was used to investigate the economics of minimum-buy policies in the U.S. Army management of spare-parts inventory. An empirical simulation makes use of historical data as direct input to the simulation rather than randomly generated data based on a characterization of the historical data. Empirical simulation alleviates the drawbacks of characterizing historical data with a theoretical distribution; however, because steady-state conditions are not reached, a methodological problem arises in the evaluation of inventory assets at the end of the simulation. Our solution was to minimize this problem by use of a cyclic approach.     

Least Absolute Value Regression: Necessary Sample Sizes to Use Normal Theory Inference Procedures*

Recently developed large sample inference procedures for least absolute value (LAV) regression are examined via Monte Carlo simulation to determine when sample sizes are large enough for the procedures to work effectively. A variety of different experimental settings were created by varying the disturbance distribution, the number of explanatory variables and the way the explanatory variables were generated. Necessary sample sizes range from as small as 20 when disturbances are normal to as large as 200 in extreme outlier-producing distributions.     

A Comparison of Three Joint Ordering Inventory Policies*

This paper evaluates the performance of a joint ordering inventory policy which was first suggested and characterized by Renberg and Planche [14]. This paper shows that the policy is easily characterized for Poisson demands. This policy is then compared with two other joint ordering policies—the well-known (S, c, s) or can-order policy of Balintfy [3] and the recent periodic policies suggested by Atkins and lyogun [2]. For a continuous review operating environment, the Renberg and Planche policy utilizes a group reorder point and a combined order quantity (Q), with each item maintaining an order-up-to level (S). For the can-order policy, each item in the product group has a must-order point (s), a can-order point (c) and an order-up-to level (S). The periodic policies require that item orders be grouped at some fixed scheduled intervals. Using long-run total average costs as the basis, it is shown that no one policy is superior to the others in all the examples tested. In some cases, the Renberg and Planche policy performs surprisingly well.     

Assessing Service Level Targets in Production and Inventory Planning*

In production and stock planning, the relationship between customer service, defined as the ability to meet demand for finished goods from in-stock inventory, and expected profits or expected costs can be represented by a simple reliability curve. The shape of this curve depends upon the parameters of the demand process, specifically the expected level of demand, standard deviation and correlation structure, as well as upon the capacities and initial state of the production and inventory system. A model is presented which explicitly determines this trade-off curve for a firm. The model is intended both as an operational model to aid managers in setting revenue and service targets which are compatible with the capacities and resources of the firm, and as a tool for exploring relationships between the parameters of the demand process and the constraints of the physical production and inventory system. The results illustrate that the level of risk depends strongly on the variability of the demand process, the cost structure, the capacities and initial state of the system and, to a lesser extent, the correlation in demand between succeeding periods. Results suggest that establishing service level targets consistent with the firm's strategic orientation must be done in consideration of both the characteristics of the demand process and the capacities of the production and inventory system. The model provides a tool for estimating the premium above unit cost which must be paid to provide a designated service level.     

Patterns of Information Source Use Across Industrial Purchase Situations*

An empirical taxonomy of industrial customers' information source use is developed based on a survey of 636 industrial customers across a wide range of different purchase situations. The taxonomy reveals five distinct information source mixes. Each mix consists of the combination of individual information sources used in a purchase situation. The five information source mixes are related to select underlying characteristics of purchase situations. The results indicate that the multivariate dimensions of purchase involvement, purchase complexity, and multiple influence are all significantly related to customers' choice of an information source mix. Implications of the taxonomy for marketing management and research are discussed.     

Integrative Cycle Scheduling Approach for a Capacitated Flexible Assembly System*

This study revisits the traditional single stage, multi-item, capacitated lot-sizing problem (CLSP) with a new integrative focus on problem structuring. Unlike past research, we develop integrative cycle scheduling approaches which simultaneously address lot-sizing, capacity, and sequencing issues. Our purposes are to (1) explore the effect of sequencing on inventory levels, (2) examine the problem of infeasibility in the economic lot scheduling problem (ELSP), and (3) provide a simple methodology of generating low-cost cycle schedules in an environment with discrete shipping, dynamic demands, limited capacity, zero setup cost, and sequence-independent setup times. Our procedures are compared to benchmark cycle scheduling approaches in terms of both inventory cost and computation time under different demand scenarios, using the operating data from a flexible assembly system (FAS) at the Ford Motor Company's Sandusky, Ohio plant.     

Inventory Positioning/Partitioning for Backorders Optimization for a Class of Multi-Echelon Inventory Problems*

Inventory management has undergone significant philosophical changes in recent decades such as the advent of the zero inventory concept. However, as attractive as the concept of minimal inventories may be, it is often unrealistic in application. Attention to basic features of inventory control systems such as order quantities, base stock levels, and reorder points remain crucial to ensure customer service at minimal cost. A nonlinear optimization model for determining base stock levels in a multi-echelon inventory network is presented. Lagrangian relaxation results in (1) newsboy-style relations that provide the optimal solutions, and (2) instantaneous shadow prices for the budget constraint. Sensitivity analysis of this model will facilitate making decisions concerning the desired investment in inventory for the entire system. This model may be solved on standard nonlinear programming software and is generalizable to problems in both production and distribution settings.     

Efficient Heuristics for MRP Lot Sizing with Variable Production/Purchasing Costs*

Two heuristics based on branch and bound (B&B) are developed to solve closed-loop material requirements planning (MRP) lot-sizing problems that have general product structures and variable costs. A “look ahead method'’(LAM) heuristic allows for variable production/purchasing costs and uses a single-level B&B procedure to rapidly improve lower bound values; thus, LAM efficiently uses computer-storage capacity and allows solution of larger problems. The “total average modification'’(TAM) heuristic uses B&B, applied level by level, and modified setup and carrying costs to solve the variable production/purchasing costs MRP lot-sizing problem. LAM and TAM are tested on problems and compared to heuristics in the literature. TAM may be used to solve large MRP lot-sizing problems encountered in practice.