Search

Same-day parcel deliveries, grocery order collection in warehouses and food delivery by bicycles are but a few recent examples that illustrate the increased complexity of the routing problems our society is facing nowadays. The increasingly complex nature of routing problems is in part caused by a tremendous growth of stochastic involvement. Travel times between customers are usually uncertain, their demands may change, and their presence at the time of delivery is often unknown. Accounting for such uncertainties plays an important role in the design of efficient routes. The body of literature dedicated to the performance of solution methods in such stochastic settings is therefore growing rapidly. However, only few studies approach the uncertainty retained by routing problems from the opposite perspective. By revisiting one of the earliest accounts in the field of stochastic routing, the probabilistic travelling salesman problem, this thesis investigates the impact of uncertainty in customer presence 'off the beaten track': How the problem characteristics of a fundamental stochastic routing problem poses challenges to the empirical performance of solution methods. The probabilistic travelling salesman problem is concerned with finding the shortest route along a set of customers in a graph, while their presence is stochastic. That is, each customer in the graph only requires a visit with a certain probability. Therefore, the subset of customers that eventually requires a visit is unknown. Since it is well-known that the optimal solution of this problem may differ from its deterministic counterpart, the travelling salesman problem, many problem-specific solution methods have been developed to cope with the stochastic nature of the probabilistic travelling salesman problem. But only little empirical research has focused on the opposite perspective: How problem characteristics affect the performance of solution methods. Starting from scratch, a first study reviews and compares solution methods for the probabilistic travelling salesman problem. It provides a comprehensive overview of the most important probabilistic methodological adaptations of exact methods, construction heuristics, improvement heuristics and metaheuristics, and discusses their strategies, key features and performance. An in-depth analysis evaluates the construction phase of a selection of several solution methods under different circumstances. The application of sufficiently explorative heuristics in the improvement phase of deterministic routing problems has led to the commonly held belief that advantages accrued by construction heuristics vanish in the final solution. Consequently, the construction phase also remained largely unexplored in the probabilistic travelling salesman problem. I put the hypothesis to the test that initial solutions found by construction heuristics contribute to the quality of final solutions for the probabilistic travelling salesman problem. Empirical research reveals that the spatial configuration of customers in the problem is an essential determinant in the extent of the contribution. Specifically, the presence of clustering of customers across the plane is one of the main drivers of the relative performance of construction heuristics. Building on this insight, a follow-up study proposes a new hyperheuristic framework that generates new construction heuristics tailored to any given problem setting. This hyperheuristic framework, named GENS-H, is the first reported application of hyperheuristics in the stochastic routing domain that explicitly takes problem characteristics into account. GENS-H relies on the Generic Savings procedure, a solution approach that unifies a class of construction heuristics, namely savings-based procedures, into a single generalised and parameterised framework. GENS embeds existing savings procedures as special instances, but is also capable of producing numerous new procedures on demand by plugging in different parameter configurations. Notwithstanding its potential, the goal of an included GENS-H benchmark study is not predominantly to outperform state-of-the-art methods, but rather to demonstrate the added value of pursuing an approach where problem characteristics - and specifically, the clustering of customers - are taken into consideration. Extensive numerical tests support this premise. The notion that spatial proximity in the form of customer clustering plays a quintessential role in the characterisation of a problem promotes the idea that dependency through other, potentially also nonspatial, interactions between customers could be an overlooked trait in models for stochastic routing. That is, the assumption made by the probabilistic travelling salesman problem that customers are independent can be regarded as an oversimplification. Drawing on concepts in computational sociology and financial mathematics, the last study of this thesis proposes a generalised version of the probabilistic travelling salesman problem that incorporates a hands-on way to model dependencies between customers who are seemingly related. This new stochastic combinatorial optimisation problem, the correlated probabilistic travelling salesman problem, ventures into an unexplored territory of stochastic routing where the probabilistic requirements to interact with customers are not isolated endeavours any more. More specifically, the correlated probabilistic travelling salesman problem integrates a heterogeneous set-up that allows one to specify correlations between the stochastic requirements of visiting a group or cluster of customers, with the ability to reduce to the original probabilistic travelling salesman problem in case dependence is absent. This implementation has many practical applications in solving real-world routing problems where, for example, similar behaviour in customer presence can be observed among different customer segments. This study also demonstrates that good solutions for the deterministic version of the travelling salesman problem and the probabilistic travelling salesman problem do not necessarily coincide with good solutions for the correlated adaptation. As such, the development of new, customised, solution methods for the correlated probabilistic travelling salesman is needed.

Since the publication of the seminal paper by Charnes, Cooper and Rhodes in 1978, where the conventional CCR model of Data Envelopment Analysis (DEA) has been proposed, DEA as a field has substantially evolved both methodologically and in terms of applications. So far, efficiency and productivity studies in the banking sector proved to be amongst the most popular application areas. The popularity of DEA in this field, amongst others, is due to its unique features such as its non-parametric nature, it benchmarks against the best practice performers rather than the average performers. It allows one to identify targets for improvement; it does not need any functional specification of the relationship between inputs and outputs, and provides a variety of efficiency measures most suitable for a variety of applications. Moreover, it provides a wide range of models to perform analyses at the aggregate level and the detailed level. In addition, DEA models allow one to perform both Static and Dynamic analyses. In this thesis, DEA is used to assess the efficiency profiles of commercial banks under heterogeneity conditions. First, a new DEA-based analysis framework with a regression-based feedback mechanism is proposed to deal with the particular features of the UK banking sector, where regression analysis provides DEA with feedback that informs about the relevance of the inputs and the outputs chosen by the analyst. Unlike previous studies, the DEA models used within the proposed framework could use both inputs and outputs, only inputs, or only outputs, which proved necessary with UK data. Second, to the best of our knowledge, no attempt has been made to investigate the relative efficiency of operating environments of banks. This thesis aims at filling this gap by analysing the efficiency of HSBC in different operating environments or countries over time. The choice of a single bank; namely, HSBC, is motivated by isolating the operating environment effect on efficiency and thus avoiding any bias that would result from the relative efficiency of different banks within the same operating environment. From a methodological perspective, this analysis is performed using a variety of framework; namely, A four-stage analysis is performed with Static black box SBM, Dynamic SBM, Network SBM, and Dynamic-Network SBM DEA frameworks. Overall, this thesis contributes to both the DEA field, through its methodological contributions, and the banking sector, through its application of the methodological contributions in assessing banks' efficiency profiles.

Combinatorial optimisation problems (COPs) have been at the origin of the design of many optimal and heuristic solution frameworks such as branch-and-bound algorithms, branch-and-cut algorithms, classical local search methods, metaheuristics, and hyperheuristics. This thesis proposes a refined generic and parametrised infeasible local search (GPILS) algorithm for solving COPs and customises it to solve the traveling salesman problem (TSP), for illustration purposes. In addition, a rule-based heuristic is proposed to initialise infeasible local search, referred to as the parameterised infeasible heuristic (PIH), which allows the analyst to have some control over the features of the infeasible solution he/she might want to start the infeasible search with. A recursive infeasible neighbourhood search (RINS) as well as a generic patching procedure to search the infeasible space are also proposed. These procedures are designed in a generic manner, so they can be adapted to any choice of parameters of the GPILS, where the set of parameters, in fact for simplicity, refers to set of parameters, components, criteria and rules. Furthermore, a hyperheuristic framework is proposed for optimizing the parameters of GPILS referred to as HH-GPILS. Experiments have been run for both sequential (i.e. simulated annealing, variable neighbourhood search, and tabu search) and parallel hyperheuristics (i.e., genetic algorithms / GAs) to empirically assess the performance of the proposed HH-GPILS in solving TSP using instances from the TSPLIB. Empirical results suggest that HH-GPILS delivers an outstanding performance. Finally, an offline learning mechanism is proposed as a seeding technique to improve the performance and speed of the proposed parallel HH-GPILS. The proposed offline learning mechanism makes use of a knowledge-base to keep track of the best performing chromosomes and their scores. Empirical results suggest that this learning mechanism is a promising technique to initialise the GA's population.

It is well established that one of the key contributing factors to the growth of countries’ economies is the existence of appropriate economic and social infrastructures. The setup of these infrastructures and their operation or delivery of public services associated with them have proved to be a costly and often unaffordable business for most governments. Therefore, private sector participation in the delivery of public services has been opted for by governments under Public-Private Partnerships (PPPs). Additional reasons for the public sector to choose a PPP framework to deliver public services include the public sector representatives’ lack of sufficient technical knowledge and specifications required by most projects; the required amount of investment to implement such projects is too high to be affordable by the public sector alone; the increasing difficulties of predicting and controlling the uncertainties that typically govern the phases of such projects including the level of actual demand for the related services by the time the project is operational; the level of bureaucracy in the public sector and its implications on the time frame of decision-making processes; the relatively low efficiency in project execution by the public sector; and most importantly the fact that, in practice, these difficulties are often better managed by the private sector who seem to possess both the financial and the executive capabilities to undertake such projects Currently, there is a gap in the game theory literature on modelling PPP related decisions and issues. This thesis contributes to reducing such gap by proposing three novel and practice-oriented game theoretical models along with algorithms for solving them, namely an ordinal game for the shortlisting of potential private sector partners; two non-cooperative dynamic games for negotiation with shortlisted private sector partners and selection of private sector partner to award the PPP contract to; and a non-cooperative dynamic game for financial renegotiation post-PPP contract award. In sum, each novel game addresses an important problem faced by the private sector during the typical stages of the life cycle of PPPs, namely private sector partner selection at the tendering stage; negotiation between the public sector and shortlisted bidders to select the candidate to whom to award a PPP contract to at the post-tendering stage; and financial re-negotiation between the public sector and the selected private sector partner at the post-contract award stage.

Prediction of corporate bankruptcy (or distress) is one of the major activities in auditing firms’ risks and uncertainties. The design of reliable models to predict distress is crucial for many decision-making processes. Although a variety of models have been designed to predict distress, the relative performance evaluation of competing prediction models remains an exercise that is unidimensional in nature. To be more specific, although some studies use several performance criteria and their measures to assess the relative performance of distress prediction models, the assessment exercise of competing prediction models is restricted to their ranking by a single measure of a single criterion at a time, which leads to reporting conflicting results. The first essay of this research overcomes this methodological issue by proposing an orientation-free super-efficiency Data Envelopment Analysis (DEA) model as a multi-criteria assessment framework. Furthermore, the study performs an exhaustive comparative analysis of the most popular bankruptcy modelling frameworks for UK data. Also, it addresses two important research questions; namely, do some modelling frameworks perform better than others by design? and to what extent the choice and/or the design of explanatory variables and their nature affect the performance of modelling frameworks? Further, using different static and dynamic statistical frameworks, this chapter proposes new Failure Prediction Models (FPMs). However, within a super-efficiency DEA framework, the reference benchmark changes from one prediction model evaluation to another one, which in some contexts might be viewed as “unfair” benchmarking. The second essay overcomes this issue by proposing a Slacks-Based Measure Context-Dependent DEA (SBM-CDEA) framework to evaluate the competing Distress Prediction Models (DPMs). Moreover, it performs an exhaustive comparative analysis of the most popular corporate distress prediction frameworks under both a single criterion and multiple criteria using data of UK firms listed on London Stock Exchange (LSE). Further, this chapter proposes new DPMs using different static and dynamic statistical frameworks. Another shortcoming of the existing studies on performance evaluation lies in the use of static frameworks to compare the performance of DPMs. The third essay overcomes this methodological issue by suggesting a dynamic multi-criteria performance assessment framework, namely, Malmquist SBM-DEA, which by design, can monitor the performance of competing prediction models over time. Further, this study proposes new static and dynamic distress prediction models. Also, the study addresses several research questions as follows; what is the effect of information on the performance of DPMs? How the out-of-sample performance of dynamic DPMs compares to the out-of-sample performance of static ones? What is the effect of the length of training sample on the performance of static and dynamic models? Which models perform better in forecasting distress during the years with Higher Distress Rate (HDR)? On feature selection, studies have used different types of information including accounting, market, macroeconomic variables and the management efficiency scores as predictors. The recently applied techniques to take into account the management efficiency of firms are two-stage models. The two-stage DPMs incorporate multiple inputs and outputs to estimate the efficiency measure of a corporation relative to the most efficient ones, in the first stage, and use the efficiency score as a predictor in the second stage. The survey of the literature reveals that most of the existing studies failed to have a comprehensive comparison between two-stage DPMs. Moreover, the choice of inputs and outputs for DEA models that estimate the efficiency measures of a company has been restricted to accounting variables and features of the company. The fourth essay adds to the current literature of two-stage DPMs in several respects. First, the study proposes to consider the decomposition of Slack-Based Measure (SBM) of efficiency into Pure Technical Efficiency (PTE), Scale Efficiency (SE), and Mix Efficiency (ME), to analyse how each of these measures individually contributes to developing distress prediction models. Second, in addition to the conventional approach of using accounting variables as inputs and outputs of DEA models to estimate the measure of management efficiency, this study uses market information variables to calculate the measure of the market efficiency of companies. Third, this research provides a comprehensive analysis of two-stage DPMs through applying different DEA models at the first stage – e.g., input-oriented vs. output oriented, radial vs. non-radial, static vs. dynamic, to compute the measures of management efficiency and market efficiency of companies; and also using dynamic and static classifier frameworks at the second stage to design new distress prediction models.

The current extremely volatile business world requires firms to deal with a wide range of risks that pose threats to their organisations. The poor practices of risk management, based on Traditional Risk Management (TRM), was cited time and time again in the aftermath of the recent Global Crisis. Enterprise Risk Management (ERM) has been advocated as a solution to the problems of TRM. The aim is to centralise the management of risk within the organisation and ensure that the board deals with the risk. Hence strategic, external, internal, operational, compliance and reputational risk are dealt with jointly. In doing so, it is expected that ERM will bring value creation to firms. One of the main limitations facing researchers is the lack of a good standardised measurement of ERM implementation; therefore, it has not been possible to establish whether ERM does actually bring benefit to firms. In addition, many companies have set up ERM initiatives, but they lack a clear understanding of the factors that will lead to successful ERM implementation. The remaining unanswered problematic situation has led to two unanswered questions that will determine whether the solution to ERM implementation is avoiding potential pitfalls and improving business sustainability. Firstly, does ERM implementation have an impact on firm performance? And secondly, which is the firm-specific characteristic that leads to better ERM implementation level? This thesis answers the aforementioned questions by proposing a reliable ERM measurement method, and then testing whether firms that adopt ERM actually improve financial performance and determine the influential factor of ERM implementation. The proposed method for measuring ERM implementation is based on the components developed from the current ERM frameworks, where contribution scoring can be standardised to measure ERM implementation level. To demonstrate its viability, data was collected from publicly listed firms in Thailand and was then compared to three alternative methodologies: cluster analysis (CA), principal component analysis (PCA) and partial least squares (PLS). The results show that the proposed method did well compared to the alternatives, both statistically and in prediction performance. The relationship between the proposed ERM measurement and firm performance is then considered by taking appropriate control variables into account, such as the firm’s size and characteristics, industry effects, sales growth and the external environment: technology, market uncertainty, as well as economic factors. By using data from the Thailand Stock Exchange, it was found that implementing ERM could improve firm performance in term of Tobin's Q, ROE and ROA. The results show that ERM and firm performance are related. For the influential factor of ERM implementation, the empirical results show that a firm’s size and economic factors have a statistically positive relationship with a high level of ERM implementation, while lower ERM scores show more revenue volatility than those who have well-implemented ERMs. Furthermore, technology and growth are positively related to each ERM in the scoring system considered.

Operations Managers of manufacturing systems, distribution systems, and supply chains address lot sizing and scheduling problems as part of their duties. These problems are concerned with decisions related to the size of orders and their schedule. In general, products share or compete for common resources and thus require coordination of their replenishment decisions whether replenishment involves manufacturing operations or not. This research is concerned with joint replenishment problems (JRPs) which are part of multi-item lot sizing and scheduling problems in manufacturing and distribution systems in single echelon/stage systems. The principal purpose of this research is to develop three new periodic review policies for stochastic joint replenishment problem. It also highlights the lack of research on joint replenishment problems with different demand classes (DSJRP). Therefore, periodic review policy is developed for this problem where the inventory system faces different demand classes that are deterministic demand and stochastic demand. Heuristic Algorithms have been developed to obtain (near) optimal parameters for the three policies as well as a heuristic algorithm has been developed for DSJRP. Numerical tests against literature benchmarks have been presented.

Interest in understanding Human Beings’ behaviour can be traced back to the early days of mankind. However, interest in consumer behaviour is relatively recent. In fact, it is only since the end of World War II and following economic prosperity of some nations (e.g., U.S.A.) that the world witnessed the rise of a new discipline in the early 1950s; namely, Marketing Research. By the end of the 1950s, academic papers on modelling and analysis of consumer behaviour started to appear (Ehrenberg, 1959; Frank, 1962). The purpose of this research is to propose an integrated decision framework for modelling consumer behaviour with respect to store incidence, category incidence, brand incidence, and size incidence. To the best of our knowledge, no published contribution integrates these decisions within the same modelling framework. In addition, the thesis proposes a new estimation method as well as a new segmentation method. These contributions aim at improving our understanding of consumer behaviour before and during consumers’ visits to the retail points of a distribution network, improving consumer behaviour prediction accuracy, and assisting with inventory management across distribution networks. The proposed modelling framework is hybrid in nature in that it uses both non-explanatory and explanatory models. To be more specific, it uses stochastic models; namely, probability distributions, to capture the intrinsic nature of consumers (i.e., inner or built-in behavioural features) as well as any unexplained similarities or differences (i.e., unobserved heterogeneity) in their intrinsic behaviour. In addition, the parameters of these probability distribution models could be estimated using explanatory models; namely, multiple regression models, such as logistic regression. Furthermore, the thesis proposes a piece-wise estimation procedure for estimating the parameters of the developed stochastic models. Also proposed is a three-step segmentation method based on the information provided by the quality of fit of stochastic models to consumer data so as to identify which model better predicts which market segments. In the empirical investigation, the proposed framework was used to study consumer behaviour with respect to individual alternatives of each decision, individual decisions, and all decisions. In addition, the proposed segmentation method was used to segment the panellists into infrequent users, light to medium users, and heavy users, on one hand, and split loyals, loyals, and hardcore loyals, on the other hand. Furthermore, the empirical evidence suggests that the proposed piece-wise estimation procedure outperforms the standard approach for all models and decision levels. Also, the empirical results revealed that the homogeneous MNL outperforms both the heterogeneous NMNL and DMNL when each one of these distributions is applied to all decisions, which suggests the relative homogeneity in consumer decision making at the aggregate or integrated decision level. Last, but not least, through the use of the proposed framework, the thesis sheds light on the importance of consumer choice sequence on the quality of predictions, which affects the quality of segmentation. The reader is referred to chapter 3 for details on these contributions.

The purpose of my research is to contribute to the field of forecasting from a methodological perspective as well as to the field of crude oil as an application area to test the performance of my methodological contributions and assess their merits. In sum, two main methodological contributions are presented. The first contribution consists of proposing a mathematical programming based approach, commonly referred to as Data Envelopment Analysis (DEA), as a multidimensional framework for relative performance evaluation of competing forecasting models or methods. As opposed to other performance measurement and evaluation frameworks, DEA allows one to identify the weaknesses of each model, as compared to the best one(s), and suggests ways to improve their overall performance. DEA is a generic framework and as such its implementation for a specific relative performance evaluation exercise requires a number of decisions to be made such as the choice of the units to be assessed, the choice of the relevant inputs and outputs to be used, and the choice of the appropriate models. In order to present and discuss how one might adapt this framework to measure and evaluate the relative performance of competing forecasting models, we first survey and classify the literature on performance criteria and their measures – including statistical tests – commonly used in evaluating and selecting forecasting models or methods. In sum, our classification will serve as a basis for the operationalisation of DEA. Finally, we test DEA performance in evaluating and selecting models to forecast crude oil prices. The second contribution consists of proposing a Multi-Criteria Decision Analysis (MCDA) based approach as a multidimensional framework for relative performance evaluation of the competing forecasting models or methods. In order to present and discuss how one might adapt such framework, we first revisit MCDA methodology, propose a revised methodological framework that consists of a sequential decision making process with feedback adjustment mechanisms, and provide guidelines as to how to operationalise it. Finally, we adapt such a methodological framework to address the problem of performance evaluation of competing forecasting models. For illustration purposes, we have chosen the forecasting of crude oil prices as an application area.