Exploring the Accuracy of Joint-Distribution Approximations Given Partial Information

Abstract

We test the accuracy of various methods for approximating underspecified joint probability distributions. In particular, we examine the maximum entropy and the analytic center approximations, and we introduce three methods for approximating a discrete joint probability distribution given partial probabilistic information. Our results suggest that recently proposed approximations and our new approximations more accurately represent the possible uncertainty models than do previous models such as maximum entropy.     

An integrated distribution routing model in multi-supply center system

A distribution routing problem with time constraint is one of the important problems in distribution and supply center management. This research is concerned with an integrated distribution routing problem for multi-supply centers based on improved genetic algorithm and graphical user interface (GUI)-type programming. In this research, we proposed a method based on a three-step approach: in step 1 a sector clustering model is developed to transfer the multi-supply center problem to single supply center problems which are easier to be solved; in step 2 we developed a vehicle routing model with time constraints and in step 3 we developed a GA-TSP model which can improve the vehicle routing schedules. The objective of the problem is to minimize the logistic cost for a set of customers without being tardy or exceeding the capacity or travel time of the vehicles. For computational purpose, we developed a GUI-type computer program according to the proposed methods and the sample outputs show that the proposed method is very effective on a set of standard test problems, and it could be potentially useful in solving the distribution routing problems.     

Discrete forecast horizons for two-product variants of the dynamic lot-size problem

Motivated by the recent success of integer programming based procedures for computing discrete forecast horizons, we consider two-product variants of the classical dynamic lot-size model. In the first variant, we impose a warehouse capacity constraint on the total ending inventory of the two products in any period. In the second variant, the two products have both individual and joint setup costs for production. To our knowledge, there are no known procedures for computing forecast horizons for these variants.Under the assumption that future demands are discrete, we characterize forecast horizons for these two variants as feasibility/optimality questions in 0–1 mixed integer programs. A detailed computational study establishes the effectiveness of our approach and enables us to gain valuable insights into the behavior of minimal forecast horizons.     

A Dynamic Double-Trigger Model of Multifamily Mortgage Default

This study advances the commercial mortgage literature by providing theory and methods for incorporating both equity and cash-flow considerations in default models. We use local market conditions to compute a (joint) probability that default is in-the-money, based on both equity and cash-flow considerations. Statistical analysis is performed using data on multifamily mortgages originated in the 1980s and early 1990s. Simulations based on statistical modeling show advantages of the probabilistic double-trigger approach over other measures of equity and cash flow.     

DERIVATION OF AN INTERTEMPORAL CORRELATION COEFFICIENT MODEL BASED ON CASH FLOW COMPONENTS AND PROBABILISTIC EVALUATION OF A PROJECT'S NPV

If the first two moments (mean and variance) of the net present value (NPV) are known, various probabilistic information of possible NPV can be derived. However, in general, it is practically impossible to estimate the variance of lengthy investment projects due to difficulties in estimating all the intertemporal correlation coefficients between cash flows of two different periods. In this paper, we derive an estimation model for the intertemporal coefficients based on cash flow components and show how the model, under a certain assumption, can be used for estimating and deriving probabilistic information.     

Measuring the efficiency of assembled printed circuit boards with undesirable outputs using data envelopment analysis

The printed circuit board (PCB) assembling production process is generally optimized to ensure very low levels of errors (defects) so as to produce the higher quality product. The evaluation of efficiency of different types of PCB (cards) could help the management in identifying the quality of assembly process to work out appropriate interventions to prevent failures. This study uses the technique of data envelopment analysis to assess the efficiency of cards across all assembly units as well as within its respective assembly unit in the presence of undesirable outputs under the assumption of weak disposability. In our case, the three types of errors, viz., machine error, manual errors and other errors that occur in the defective assembled PCBs are considered as the undesirable outputs. In-depth analysis has been carried out through jackknifing analysis to check the robustness of DEA efficiency scores. The best practice group of cards for each inefficient card has been identified which eventually facilitate the management to derive the level of undesirable outputs to be reduced and desirable output to be improved so as to achieve the level of efficiency at par with its practice group.     

Can market power really be estimated?

In a previous paper in this Review, Hyde and Perloff ask the question, “Can Market Power be Estimated?” using the “structural model” or two proposed alternatives. They find that none of the three methods produce consistent, meaningful results in the food and beverage industries and that simulations suggest great sensitivity of results to model misspecification. This paper offers an explanation for these disappointing results. The difficulty is that the structural model (and its proposed alternatives) is based on an overly-ambitious estimating form that over-simplifies the diversity and true complexity of oligopoly pricing. By neglecting accounting data on costs and by arbitrarily relying on an assumption that outputs are set by subtracting a constant proportion of the difference between Price and Marginal Revenue, the “structural model” and proposed alternatives try to fit an elegant form to a messy world with predictably disappointing results.     

天然气管道裂纹和材料性能参量的概率统计分析

对某天然气管道的裂纹和有关材料性能参量进行了试验测定和试验数据的数理统计处理，建立了管线有关参量的统计分布函数的概率模型。对该管道主要随机变量的统计分析表明，裂纹深度和材料的断裂韧性表现为威布尔分布．裂纹深长比表现为对数正态分布．材料的屈服强度和拉伸强度表现为正态分布。     

Unionization and Job Design Under Programmable Automation

New findings from an original national survey indicate that machining jobs include major computer programming responsibilities far more commonly than received labor process theory would lead us to expect or predict. In unionized plants, workers are less likely to program their machines, perhaps because of management's desire to avoid the constraints imposed by collective bargaining agreements. Among recent adopters of the technology, informal bargaining through joint labor-management problem-solving committees neutralizes this negative union effect.     

Addressing the advantages of using ensemble probabilistic models in Estimation of Distribution Algorithms for scheduling problems

Estimation of Distribution Algorithms (EDAs) have recently been recognized as a prominent alternative to traditional evolutionary algorithms due to their increasing popularity. The core of EDAs is a probabilistic model which directly impacts performance of the algorithm. Previous EDAs have used a univariate, bi-variate, or multi-variable probabilistic model each time. However, application of only one probabilistic model may not represent the parental distribution well. This paper advocates the importance of using ensemble probabilistic models in EDAs. We combine the univariate probabilistic model with the bi-variate probabilistic model which learns different population characteristics. To explain how to employ the two probabilistic models, we proposed the Ensemble Self-Guided Genetic Algorithm (eSGGA). The extensive computation results on two NP-hard scheduling problems indicate the advantages of adopting two probabilistic models. Most important of all, eSGGA can avoid the computation effort overhead when compared with other EDAs employing two models. As a result, this paper might point out a next generation approach for EDAs.     

A case study of condition based maintenance modelling based upon the oil analysis data of marine diesel engines using stochastic filtering

This paper presents a case study of condition based maintenance modelling based on measured metal concentrations observed in oil samples of a fleet of marine diesel engines. The decision model for optimising the replacement time of the diesel engines conditional on observed measurements is derived and applied to the case discussed. We described the datasets, which were cleaned and re-organised according to the need of the research. The residual time distribution required in the decision model was formulated using a technique called stochastic filtering. Procedures for model parameter estimation are constructed and discussed in detail. The residual life model presented has been fitted to the case data, and the modelling outputs are discussed.     

Multiproduct Outputs and Scale Economies in Electric Power Production: Some New Estimates

The present study offers an argument, through a technical exposition, which suggests that there are cost differences inherent in serving different types of buyers in the electric utilities industry. To this end, we employ a translog joint cost function (for electricity production) with three outputs, which allows for the examination of the cost differences between wholesale and retail trade for vertically integrated, investor-owned companies in the industry. The regression estimates suggest different costs at the means, however the estimated marginal cost curve for wholesale power is not as robust as those for low-voltage and industrial power.     

A Comparison of Fuzzy Set and Probabilistic Paradigms for Ranking Vague Economic Investment Information Using a Present Worth Criterion

This article introduces an approach for comparing the fuzzy set and probabilistic paradigms for ranking vague economic investment information when a present worth criterion is used. Comparisons are made between the two paradigms by forming a ratio of fuzzy ranking value to probabilistic ranking value for a set of tested economic decision conditions (vague cash flows and interest rates). A multivariable analysis of the comparison ratios indicates the dominance or spread required in the expected present worths of competing alternatives to guarantee consistent decision ranking among the paradigms. Results of the research suggest that probabilistic methods will not rank vague economic investment information the same as fuzzy set methods. As a result, decision makers who use probabilistic methods to model vagueness in human thought (the condition for which fuzzy theory was developed) may obtain rankings inconsistent with those obtained via the fuzzy models. In such cases, if decisions are based on project rankings, then different choices or recommendations are likely.     

基于DEA的我国钢铁企业节能减排潜力研究

本文基于数据包络分析(DEA)的基础理论和方法,从不期望产出的角度出发,将钢铁产业污染物排放作为一种非期望输入引入到数据包络分析模型中,并将该模型应用于我国钢铁产业生态效率评价及节能减排潜力的研究。研究结论表明,我国各钢铁企业生态效率总体上不高,但呈现出逐渐增强的趋势;通过进一步的钢铁企业效率改进分析得知,我国钢铁企业还存在着较严重的投入拥挤和产出不足现象,仅从钢铁企业的47家样本来看,我国钢铁企业有着巨大的节能减排潜力。     

Immiserising invention : The private and social returns to R&D under oligopoly

A cost-saving invention may so enhance a firm's market power that output price rises. The private return to R&D may then exceed the realised social gain. These possibilities arise when integer constraints are important in determining industry size. If economies of scale are exhausted at sufficiently low outputs for free entry to result in a zero-profit equilibrium, small inventions leave output price unchanged. This mirrors Arrow's perfectly competitive model. But under oligopoly both fixed fees and (possibly negative) royalties are used by an inventor in earning a return which may exceed the potential social gain.     

An investment model of durable good diffusion

Existing studies that deal with the diffusion of durable good innovations have been criticized for their lack of an explicit testable theory of new product growth. This paper attempts to remedy this situation by providing a theoretical model of market penetration of new durable goods derived from The basic assumption that potential users of the new intermediate product attempt to minimize costs. The resulting model defines a time path of short-run equilibrium market shares determined by the cost characteristics of both the new innovation and the equipment that it is designed to replace, the age distribution of the existing capital stock, and the growth rate of the adopting sector.     

Netzausbauplanung unter Berücksichtigung probabilistischer Einflussgrößen

The present network expansion planning is performed by simulating extreme scenarios in order to proof the security of energy supply. This procedure is associated with high investment costs for equipment which is only needed a few hours per year. Planners have to consider uncertainties and prediction errors for the creation of the extreme scenarios due to the long economic life time of the equipment. The consideration of these influencing factors becomes very complex due to increased utilization of renewable energy sources and frequent changes in the unit commitment of conventional power plants. In contrast to classic load flow methods probabilistic load flow calculation considers the static behaviour of the loads, renewable energies and the power plants instead of only discrete values of nodal powers. These characteristics can be described by probabilistic density functions of load and generation mapped with a probabilistic load flow calculation into a probabilistic density function of the state variables of the energy system (as nodal voltages or line currents). This article presents an overview about existing methods of probabilistic load flow calculations and introduces a newly developed method. The characteristics and the applicability of these methods are verified and the computational burden is analysed exemplarily. Necessary extensions for the mapping of real energy systems and possible approaches for this are presented. The potential and the impact of probabilistic network expansion planning are discussed.     

The theory and practice of workload control: A research agenda and implementation strategy

The Workload Control (WLC) concept is one of few Production Planning and Control (PPC) solutions appropriate for Make-To-Order (MTO) companies yet its successful implementation is an enduring challenge. Most implementations reported are in large organisations yet it has been argued that WLC is particularly suitable for Small and Medium sized Enterprises (SMEs) with limited financial resources. Moreover, previous studies do not adequately describe the process through which implementation success was achieved. In this paper, data collected through semi-structured face-to-face interviews with key personnel from 41 companies is presented in order to build up a body of evidence on the characteristics of MTO SMEs that affect WLC implementation. The data paints a complex picture of MTO production and suggests that research is likely to be required to move theory closer to practice and organisational change to move practice closer to theory. The former is recommended where processes perform well or appear difficult to change; the latter is recommended where fundamental WLC principles are involved. Two outputs emerge from the study: (1) a research agenda for refining the WLC concept to improve alignment between theory and practice, covering the customer enquiry, order entry, design and engineering and order release stages; and (2) an implementation strategy, including elements of organisational change and covering pre-implementation, implementation and post-implementation stages. The characteristics of MTO SMEs identified should be used in future research to develop more realistic simulations for testing conceptual refinements while field research should apply and extend the implementation strategy presented in order to develop a more detailed roadmap for successful WLC implementation in practice.     

Buyers' perspectives of buyer-seller relationship development

Long-term buyer-seller relationships have been a focus of research for several years. The present study draws on interaction/network theory to test a model examining the interaction mechanisms and relationship characteristics of buyer-seller relationships during four progressive phases of relationship development. Data from 174 members of the Institute for Supply Management offer empirical support for the associations proposed in the model, as well as some surprising results. One of the most useful findings of the research is that the patterns of these associations vary as buyer-seller relationships progress through the four phases of relationship development. Specifically: in the awareness phase, joint problem solving increases buyer uncertainty; in the exploration phase, communication quality and joint problem solving increase relationship-specific investments; and in the expansion phase, joint problem solving increases relationship-specific investments and severe conflict resolution increases buyer uncertainty. Seller reputation moderates many of these relationships. The major conclusion of the research is that buyers and sellers should recognize that while information exchange and conflict resolution are important aspects of buyer-seller relationships, their use may not always lead to the desired relationship characteristics. Managerial implications of these findings and further research ideas are presented.     

A methodology for optimal product positioning with engineering constraints consideration

Due to the complex mapping relationship between product attributes and engineering characteristics, as well as the correlations between engineering characteristics and the engineering constraints on a product, a new product following a probabilistic rule of multidimensional scaling may not be in an optimal position in product engineering space. In this paper, a new methodology for optimal product positioning by considering engineering constraints is proposed. In the proposed methodology, perceptual mapping and house of quality are introduced to link the consumer perceptual space, and product engineering space. The degree of overall customer satisfaction is considered in the rule of consumer choice probability. Based on this, an optimal product positioning model can be established. Genetic algorithms are introduced to solve the problem of the optimization model due to its non-linear characteristics. By applying genetic algorithms, the optimal value settings of a new product's engineering characteristics can be obtained. An example of optimal positioning and determination of value settings of engineering characteristics of packing machines is used to illustrate the proposed methodology.