Operational Research Commons^™

Application Of Bayesian Networks In Consumer Service Industry, Yuan Gao

Open Access Theses

Gao, Yuan. M.S.I.E., Purdue University. December 2014. Application of Bayesian Networks in Consumer Service Industry. Major professor: Vincent G. Duffy The purpose of the present study is to explore the application of Bayesian networks in the consumer service industry to model causal relationships within complex risk factor structures using aggregate data. An analysis of the Hawaii tourism market was conducted to find out how visitor characteristics affect their behavior and experience as consumers during the trips, and influence the tourism market outcomes represented by measurable factors. Two hypotheses were proposed regarding the use of aggregate data and the influence of …

Location And Capacity Modeling Of Network Interchanges, Aldo D. Fabregas

USF Tampa Graduate Theses and Dissertations

Network design decisions, especially those pertaining to urban infrastructure, are made by a central authority or network leader, and taking into consideration the network users or followers. These network decision problems are formulated as non-linear bi-level programming problems. In this work, a continuous network design problem (CNDP) and discrete network design problem (DNDP) bi-level optimization programs are proposed and solved in the context of transportation planning. The solution strategy involved reformulation and linearization as a single-level program by introducing the optimality conditions of the lower level problem into the upper level problem. For the CNDP, an alternative linearization algorithm (modified …

Meta-Heuristics Analysis For Technologically Complex Programs: Understanding The Impact Of Total Constraints For Schedule, Quality And Cost, Henry Darrel Webb

EMSE Doctoral Projects

Program management data associated with a technically complex radio frequency electronics base communication system has been collected and analyzed to identify heuristics which may be utilized in addition to existing processes and procedures to provide indicators that a program is trending to failure. Analysis of the collected data includes detailed schedule analysis, detailed earned value management analysis and defect analysis within the framework of a Firm Fixed Price (FFP) incentive fee contract.

This project develops heuristics and provides recommendations for analysis of complex project management efforts such as those discussed herein. The analysis of the effects of the constraints on …

Self Learning Strategies For Experimental Design And Response Surface Optimization, Adel Alaeddini

Wayne State University Dissertations

Most preset RSM designs offer ease of implementation and good performance over a wide range of process and design optimization applications. These designs often lack the ability to adapt the design based on the characteristics of application and experimental space so as to reduce the number of experiments necessary. Hence, they are not cost effective for applications where the cost of experimentation is high or when the experimentation resources are limited. In this dissertation, we present a number of self-learning strategies for optimization of different types of response surfaces for industrial experiments with noise, high experimentation cost, and requiring high …

Investigating An Optimal Decision Point For Probability Bounds Analysis Models When Used To Estimate Remedial Soil Volumes Under Uncertainty At Hazardous Waste Sites, Charles O. Dankwah

Walden Dissertations and Doctoral Studies

Hazardous waste site remediation cost estimation requires a good estimate of the contaminated soil volume. The United States Environmental Protection Agency (U.S. EPA) currently uses deterministic point values to estimate soil volumes but the literature suggests that probability bounds analysis (PBA) is the more accurate method to make estimates under uncertainty. The underlying statistical theory is that they are more accurate than deterministic estimates because probabilistic estimates account for data uncertainties. However, the literature does not address the problem of selecting an optimal decision point from the interval-valued PBA estimates. The purpose of this study was to identify the optimal …

A Confidence Paradigm For Classification Systems, Nathan J. Leap

Theses and Dissertations

There is no universally accepted methodology to determine how much confidence one should have in a classifier output. This research proposes a framework to determine the level of confidence in an indication from a classifier system where the output is or can be transformed into a posterior probability estimate. This is a theoretical framework that attempts to unite the viewpoints of the classification system developer (or engineer) and the classification system user (or war-fighter). The paradigm is based on the assumptions that the system confidence acts like, or can be modeled as a value and that indication confidence can be …

Improving Mixed Variable Optimization Of Computational And Model Parameters Using Multiple Surrogate Functions, David Bethea

Theses and Dissertations

This research focuses on reducing computational time in parameter optimization by using multiple surrogates and subprocess CPU times without compromising the quality of the results. This is motivated by applications that have objective functions with expensive computational times at high fidelity solutions. Applying, matching, and tuning optimization techniques at an algorithm level can reduce the time spent on unprofitable computations for parameter optimization. The objective is to recover known parameters of a flow property reference image by comparing to a template image that comes from a computational fluid dynamics simulation, followed by a numerical image registration and comparison process. Mixed …

Radical Impact Of Change In Actions And Confidence Index On Reverse Decision Making An Application Based Study, Swatee Trimbak Paithankar

Engineering Management & Systems Engineering Theses & Dissertations

While making decisions under uncertainty, people are often unaware of the logical approach to form the decision process. It is assumed that collecting details, analyzing and evaluating data is enough to make 'proper' decisions. However, past research in the decision making arena has significantly validated that there exists a class of decision problems which is complex, ill-structured and not defined to the level where decision makers can draw logical conclusions based on existing traditional decision approaches. RDM (reverse decision making), one of the novel approaches of decision making under conditions of uncertainty, has shown potential towards addressing some of these …

Choice And Optimization Of Forecasting Models For Container Port Throughput, Qingcheng Xue

World Maritime University Dissertations

No abstract provided.

Uncertainty Assessment In High-Risk Environments Using Probability, Evidence Theory And Expert Judgment Elicitation, Stella Barberis Bondi

Engineering Management & Systems Engineering Theses & Dissertations

The level of uncertainty in advanced system design is assessed by comparing the results of expert judgment elicitation to probability and evidence theory. This research shows how one type of monotone measure, namely Dempster-Shafer Theory of Evidence can expand the framework of uncertainty to provide decision makers a more robust solution space. The issues imbedded in this research are focused on how the relevant predictive uncertainty produced by similar action is measured.

This methodology uses the established approach from traditional probability theory and Dempster-Shafer evidence theory to combine two classes of uncertainty, aleatory and epistemic. Probability theory provides the mathematical …

A Multivariate Magnitude Robust Control Chart For Mean Shift Detection And Change Point Estimation, Ryan M. Harrell

Theses and Dissertations

Statistical control charts are often used to detect a change in an otherwise stable process. This process may contain several variables affecting process stability. The goal of any control chart is to detect an out-of-control state quickly and provide insight on when the process actually changed. This reduces the off-line time the quality engineer spends assigning causality. In this research, a multivariate magnitude robust chart (MMRC) was developed using a change point model and a likelihood-ratio approach. Here the process is considered in-control until one or more normally distributed process variables permanently and suddenly shifts to out-of-control, stable value. Using …

Improved Hyperspectral Image Testing Using Synthetic Imagery And Factorial Designed Experiments, Joseph P. Bellucci

Theses and Dissertations

The goal of any remote sensing system is to gather data about the geography it is imaging. In order to gain knowledge of the earth's landscape, post-processing algorithms are developed to extract information from the collected data. The algorithms can be intended to classify the various ground covers in a scene, identify specific targets of interest, or detect anomalies in an image. After the design of an algorithm comes the difficult task of testing and evaluating its performance. Traditionally, algorithms are tested using sets of extensively ground truthed test images. However, the lack of well characterized test data sets and …

Optimization Of A Multi-Echelon Repair System Via Generalized Pattern Search With Ranking And Selection: A Computational Study, Derek D. Tharaldson

Theses and Dissertations

With increasing developments in computer technology and available software, simulation is becoming a widely used tool to model, analyze, and improve a real world system or process. However, simulation in itself is not an optimization approach. Common optimization procedures require either an explicit mathematical formulation or numerous function evaluations at improving iterative points. Mathematical formulation is generally impossible for problems where simulation is relevant, which are characteristically the types of problems that arise in practical applications. Further complicating matters is the variability in the simulation response which can cause problems in iterative techniques using the simulation model as a function …

An Investigation Of The Effects Of Correlation, Autocorrelation, And Sample Size In Classifier Fusion, Nathan J. Leap

Theses and Dissertations

This thesis extends the research found in Storm, Bauer, and Oxley, 2003. Data correlation effects and sample size effects on three classifier fusion techniques and one data fusion technique were investigated. Identification System Operating Characteristic Fusion (Haspert, 2000), the Receiver Operating Characteristic Within Fusion method (Oxley and Bauer, 2002), and a Probabilistic Neural Network were the three classifier fusion techniques; a Generalized Regression Neural Network was the data fusion technique. Correlation was injected into the data set both within a feature set (autocorrelation) and across feature sets for a variety of classification problems, and sample size was varied throughout. Total …

Minimum Distance Estimation For Time Series Analysis With Little Data, Hakan Tekin

Theses and Dissertations

Minimum distance estimate is a statistical parameter estimate technique that selects model parameters that minimize a good-of-fit statistic. Minimum distance estimation has been demonstrated better standard approaches, including maximum likelihood estimators and least squares, in estimating statistical distribution parameters with very small data sets. This research applies minimum distance estimation to the task of making time series predictions with very few historical observations. In a Monte Carlo analysis, we test a variety of distance measures and report the results based on many different criteria. Our analysis tests the robustness of the approach by testing its ability to make predictions when …

Computer-Based Methods For Constructing Two-Level Fractional-Factorial Experimental Designs With A Requirement Set, Steven L. Forsythe

Theses and Dissertations

This dissertation developed four methodologies for computer-aided experimental design of two-level fractional factorial designs with requirement sets (DOE/RS). The requirement sets identify all the experimental factors and the appropriate interaction terms to be evaluated in the experiment. Taguchi graphs and similar manual methods provide techniques for solving the DOE/RS problem. Unfortunately, these methods are limited because they become difficult to use as the number of factors or interaction terms exceeds ten. This research showed that the DOE/RS problem belongs to a class of difficult-to-solve problems known as NP-Complete. It is the combinatorial nature of NP-Complete problems that causes them to …

A New Sequential Goodness Of Fit Test For The Three-Parameter Gamma Distribution With Known Shape Based On Skewness And Kurtosis, Chil Ho Park

Theses and Dissertations

This research presents a new sequential goodness of fit test for the three-parameter gamma distribution with a known shape. The test is accomplished by employing two new tests, sample skewness and sample kurtosis, sequentially as test statistics. Unlike the typical goodness of fit test, using parameter estimation methods such as maximum likelihood estimation and minimum distance estimation, this test using the two test statistics above does not involve a substantial degree of computational complexity. Large Monte Carlo simulation has been used to determine critical values and overall significance levels for all combinations of the two tests, and to conduct extensive …

A Comparison Of Circular Error Probable Estimators For Small Samples, Charles E. Williams

Theses and Dissertations

Several previous studies investigated the performance of competing circular error probable (CEP) estimators for small samples. This estimation is important in ICBM analysis because, due to expense, there are a limited number of ICBM test launches. In the most recent previous study (1993), Tongue considered five CEP estimators in a simulation test, attempting to determine the behavior of these estimators for populations of various bias, ellipticity, correlation, and sample size. In this paper, we build on Tongue's findings in three ways: (1) The number of estimators compared is expanded to eight. (2) Different factors and factor levels are used. (3) …

Experiments In Aggregating Air Ordnance Effectiveness Data For The Tacwar Model, James E. Parker

Theses and Dissertations

An interactive MS Access&trademark; based application that aggregates the output of the SABSEL model for input into the TACWAR model is developed. The application was developed following efforts to create a functional approximation of the SABSEL data using neural networks, statistical networks, and traditional statistical techniques. These approximations were compared to a look-up table methodology on the basis of accuracy, (RMSE

A Robust Method Of Solving Nonlinear Boundary Value Problems Via Modified Compromise Programming, John L. Zornick

Theses and Dissertations

This study is an extension of Ng's previous work in which goal programming was used to determine an approximate solution to a boundary value problem. This approach follows the same basic approach developed by Ng in which the method of collocation was recast as a compromise programming model. Hence, instead of solving a system of simultaneous nonlinear equations, one seeks a compromise solution which minimizes (in a weighted residual sense) a vector norm of the differential equation residuals. A difference in this approach is that it makes use of a genetic algorithm as the optimizing engine as opposed to the …

Predicting Utility Bills For Air Combat Command A Study Of Forecasting Techniques, William L. Luthie

Engineering Management & Systems Engineering Theses & Dissertations

Many companies use forecasting techniques as a tool in managing their assets. Trends in such items as sales, population and inventory levels have all been determined at one time or another using forecasting, yet research indicates that these tools have not been utilized to predict utility budgets. This research was conducted to determine if such techniques could be applied to the specific task of predicting the utility bill at an Air Force base. Three quantitative models were chosen, the Moving Average, Exponential Smoothing and Regression, to determine their applicability to the task at hand. One base within Air Combat Command, …

Estimation Of The Captive-Carry Survival Function For The Advanced Medium Range Air-To-Air Missile (Amraam), David R. Denhard

Theses and Dissertations

This thesis considers the problem of estimating the survival function of an item (probability that the item functions for a time greater than a given time t) from sampling data subject to partial right censoring (a portion of the items in the sampling data have not yet been observed to fail). Specifically the thesis describes several parametric and non-parametric statistical models that can be used when the sampling data is subject to partial right censoring. These models are applied to the case of estimating the captive-carry survival function of the AIM-120A Advanced Medium Range Air-to-Air Missile (AMRAAM).

A New Goodness-Of-Fit Test For The Gamma Distribution Based On Sample Spacings From Complete And Censored Samples, Huseyin Duman

Theses and Dissertations

This thesis studies a new goodness-of-fit test for the gamma distribution with known shape parameter. This test statistic, Z*, is based on spacings from complete or censored samples. The size of samples varied between 5 and 35. The critical value tables were generated for the Z* test statistic for complete and censored samples. The critical values were obtained for five different significance levels: 0.20 0.15, 0.10, 0.05, and 0.01. An extensive power study, containing 50,000 Monte Carlo runs was conducted using nine alternative distributions, H_a. It was observed that the Z* test statistic was more powerful against certain …

Proactive Monitoring Of Performance In Stochastic Communication Networks, John C. C. Van Hove

Theses and Dissertations

This research proposes several models for placing bounds on the expected values of some dynamic performance measures for computer communication networks with failing components. These models provide an understanding of expected network performance that is useful in the process of proactive performance monitoring and also in defining level of service agreements with network users. There were three objectives for this research. The first objective was to extend some existing models of steady-state stochastic network performance to a dynamic network flow representation in order to capture the dynamic nature of proactive monitoring. The second objective was to convert the extended absolute …

A Parametric Regression Of The Cost Of Base Realignment Action (Cobra) Model, Douglas D. Hardman, Michael S. Nelson

Theses and Dissertations

This study develops a parametric model that is capable of generating accurate estimates of the costs to close Air Force installations. The new model is based upon, but much simpler to use than, the Cost of Base Realignment Action (COBRA) model. COBRA is an economic cost analysis model that requires a minimum of 250 inputs and as many as 700 inputs. The new parametric model requires just 10 input variables and was developed using least squares multiple regression. Comparison of the new parametric model to COBRA indicates that it captures 91 percent of the variance in cost estimates generated by …

A Comparison Of Variable Selection Criteria For Multiple Linear Regression: A Second Simulation Study, David P. Woollard

Theses and Dissertations

This thesis implements a variable selection method proposed by Alan J. Miller, and makes an extension of Ross J. Hansen's 1988 thesis research by comparing the methods he examined: Minimum MSE, Minimum Sp, and Minimum Cp with Miller's method. Response Surface methodology is employed with two performance measures: the percentage of correct variables in a model and the Theoretical Mean Squared Error of Prediction (TMSEP). Each technique is applied on generated data with known multicollinearities, variances, random predictors, and sample sizes. Both performance measures are computed for models selected under each technique. A full factorial design using each performance measure …

Modified Anderson-Darling And Cramer-Von Mises Goodness-Of-Fit Tests For The Normal Distribution, David A. Gwinn Sr.

Theses and Dissertations

New techniques for calculating goodness-of-fit statistics for normal distributions with parameters estimated from the sample are investigated. Samples are generated for a Normal(0,1) distribution. Critical values are calculated for five modifications to the Anderson-Darling statistic and five modifications to the Cramer-Von Mises statistic. An extensive power study is done to test the power of the new statistics versus the power of the unmodified statistics. Powers of six of the new statistics show minimal to no improvement, two of the new statistics show a marked decrease in power, and two of the new statistics show an overall increase in power over …

A Modified Anderson Darling Goodness-Of-Fit Test For The Gamma Distribution With Unknown Scale And Location Parameters, Tamer Ozmen

Theses and Dissertations

A new modified Anderson-Darling goodness-of-fit test is introduced for the three-parameter Gamma distribution when the location parameter is found by minimum distance estimation and scale parameter by maximum likelihood estimation. Monte Carlo simulation studies were performed to calculate the critical values for A-D test when A-D statistic is minimized. These critical values are then used for testing whether a set of observations follows a Gamma distribution when the scale and location parameters axe unspecified and are estimated from the sample. Functional relationship between the critical values of A-D is also examined for each shape parameter by the variables, sample size …

A Modified Chi-Squared Goodness-Of-Fit Test For The Three-Parameter Gamma Distribution With Unknown Parameters, Thomas J. Sterle

Theses and Dissertations

A modified chi-squared goodness-of-fit test was created for the gamma distribution in the case where all three parameters must be estimated from the sample. Critical values are generated using a Monte Carlo simulation procedure with 5000 repetitions each. Random samples of 8 different sizes were drawn from gamma distributions with shape parameters 1, 1.5, 2., and 2.5. The shape, scale, and location parameters were then estimated from each sample, using an iterative technique combining the maximum likelihood and minimum distance methods, enabling, computation of the chi-squared statistics and critical values. The same process is used to generate random samples, parameter …

A New Goodness-Of-Fit Test For The Weibull Distribution Based On Spacings, Mark C. Coppa

Theses and Dissertations

The critical values for a new goodness-of-fit test based on spacings are generated for the Weibull distribution when the shape parameter is known. The critical values are used for testing whether a set of observations follow a Weibull distribution when the scale and location parameters are unknown. A Monte Carlo simulation with 10,000 iterations is used to generate the critical values for sample sizes 5(5)35 at shape parameters k equal to 0.5(0.5)1.5 and for sample sizes 5(5)20 at shape parameters k = 2.0(1.0)4.0. A Monte Carlo power study of the Z* test statistic using 5000 iterations is accomplished using nine …