Structural equation modelling (SEM) is carried out with the aim of testing hypotheses on the model of the researcher in a quantitative way, using the sampled data. Although SEM has developed in many aspects over the past few decades, there are still numerous advances which can make SEM an even more powerful technique. We propose representing the nal theoretical SEM by a Bayesian Network (BN), which we would like to call a Probabilistic Structural Equation Model (PSEM). With the PSEM, we can take things a step further and conduct inference by explicitly entering evidence into the network and performing di erent types of inferences. Because the direction of the inference is not an issue, various scenarios can be simulated using the BN. The augmentation of SEM with BN provides signi cant contributions to the eld. Firstly, structural learning can mine data for additional causal information which is not necessarily clear when hypothesising causality from theory. Secondly, the inference ability of the BN provides not only insight as mentioned before, but acts as an interactive tool as the `what-if' analysis is dynamic. / Mini Dissertation (MCom)--University of Pretoria, 2018. / Statistics / MCom / Unrestricted

Yes / Flaring is a combustion process commonly used in the oil and gas industry to dispose flammable waste gases. Flare flameout occurs when these gases escape unburnt from the flare tip causing the discharge of flammable and/or toxic vapor clouds. The toxic gases released during this process have the potential to initiate safety hazards and cause serious harm to the ecosystem and human health. Flare flameout could be caused by environmental conditions, equipment failure, and human error. However, to better understand the causes of flare flameout, a rigorous analysis of the behavior of flare systems under failure conditions is required. In this article, we used fault tree analysis (FTA) and the dynamic Bayesian network (DBN) to assess the reliability of flare systems. In this study, we analyzed 40 different combinations of basic events that can cause flare flameout to determine the event with the highest impact on system failure. In the quantitative analysis, we use both constant and time-dependent failure rates of system components. The results show that combining these two approaches allows for robust probabilistic reasoning on flare system reliability, which can help improving the safety and asset integrity of process facilities. The proposed DBN model constitutes a significant step to improve the safety and reliability of flare systems in the oil and gas industry.

No description available.

The field of medical informatics comprises many subdisciplines, united by a common interest in the establishment of standards to facilitate the sharing, reuse, and understanding of information. This work depends in large part on the ability of controlled medical terminologies to represent relevant concepts. This work augments a controlled terminology to provide not only standardized content, but also standardized explanatory knowledge for use in expert systems. This experiment consisted of four phases centered on the use of the controlled terminology-- Systemized Nomenclature of Medicine (SNOMED). The first phase evaluated SNOMED's ability to express explanatory knowledge for clinical pathology. The second developed the Normalized Medical Explanation (NORMEX) syntax for expressing and storing pathways of causal reasoning in the domain of clinical pathology. The third segment examined SNOMED's capacity to represent concepts used in the NORMEX model of clinical pathology. The final phase incorporated NORMEX-based pathways of influence in a Bayesian network to assess ability to predict causal mechanisms as implied by serum analyte results. Findings from this work suggest that SNOMED's capacity to represent explanatory information parallels its coverage of clinical pathology findings. However, SNOMED currently lacks much of the content necessary for both of these purposes. Additional explanatory content was created with an ontology-modeling tool. The NORMEX syntax was defined by SNOMED hierarchy names. Complex sequences of explanations were created using the NORMEX syntax. In addition, medical explanatory knowledge represented in the NORMEX format could be stored in an architectural framework consistent with that used by a controlled terminology such as SNOMED. Once stored, such knowledge could be retrieved from storage without loss of meaning or introduction of errors. Lastly, a Bayesian network constructed from the retrieved NORMEX knowledge produced a network whose prediction performance equaled or exceeded that of a network produced by more traditional means. / Ph. D.

<p dir="ltr">In recent years, probabilistic graphical models have emerged as a powerful framework for understanding complex dependencies in multivariate data, offering a structured approach to tackle uncertainty and model complexity. These models have revolutionized the way we interpret the interplay between variables in various domains, from genetics to social network analysis. Inspired by the potential of probabilistic graphical models to provide insightful data analysis while addressing the challenges of high-dimensionality and computational efficiency, this dissertation introduces two novel methodologies that leverage the strengths of graphical models in high-dimensional settings. By integrating advanced inference techniques and exploiting the structural advantages of graphical models, we demonstrate how these approaches can efficiently decode complex data patterns, offering significant improvements over traditional methods. This work not only contributes to the theoretical advancements in the field of statistical data analysis but also provides practical solutions to real-world problems characterized by large-scale, complex datasets.</p><p dir="ltr">Firstly, we introduce a novel Bayesian hybrid method for learning the structure of Gaus- sian Bayesian Networks (GBNs), addressing the critical challenge of order determination in constraint-based and score-based methodologies. By integrating a permutation matrix within the likelihood function, we propose a technique that remains invariant to data shuffling, thereby overcoming the limitations of traditional approaches. Utilizing Cholesky decompo- sition, we reparameterize the log-likelihood function to facilitate the identification of the parent-child relationship among nodes without relying on the faithfulness assumption. This method efficiently manages the permutation matrix to optimize for the sparsest Cholesky factor, leveraging the Bayesian Information Criterion (BIC) for model selection. Theoretical analysis and extensive simulations demonstrate the superiority of our method in terms of precision, recall, and F1-score across various network complexities and sample sizes. Specifically, our approach shows significant advantages in small-n-large-p scenarios, outperforming existing methods in detecting complex network structures with limited data. Real-world applications on datasets such as ECOLI70, ARTH150, MAGIC-IRRI, and MAGIC-NIAB further validate the effectiveness and robustness of our proposed method. Our findings contribute to the field of Bayesian network structure learning by providing a scalable, efficient, and reliable tool for modeling high-dimensional data structures.</p><p dir="ltr">Secondly, we introduce a Bayesian methodology tailored for Gaussian Graphical Models (GGMs) that bridges the gap between GBNs and GGMs. Utilizing the Cholesky decomposition, we establish a novel connection that leverages estimated GBN structures to accurately recover and estimate GGMs. This innovative approach benefits from a theoretical foundation provided by a theorem that connects sparse priors on Cholesky factors with the sparsity of the precision matrix, facilitating effective structure recovery in GGMs. To assess the efficacy of our proposed method, we conduct comprehensive simulations on AR2 and circle graph models, comparing its performance with renowned algorithms such as GLASSO, CLIME, and SPACE across various dimensions. Our evaluation, based on metrics like estimation ac- curacy and selection correctness, unequivocally demonstrates the superiority of our approach in accurately identifying the intrinsic graph structure. The empirical results underscore the robustness and scalability of our method, underscoring its potential as an indispensable tool for statistical data analysis, especially in the context of complex datasets.</p>

Yes / The offshore wind energy is increasingly becoming an attractive source of energy due to having lower environmental impact. Effective operation and maintenance that ensures the maximum availability of the energy generation process using offshore facilities and minimal production cost are two key factors to improve the competitiveness of this energy source over other traditional sources of energy. Condition monitoring systems are widely used for health management of offshore wind farms to have improved operation and maintenance. Reliability of the wind farms are increasingly being evaluated to aid in the maintenance process and thereby to improve the availability of the farms. However, much of the reliability analysis is performed offline based on statistical data. In this article, we propose a drone-assisted monitoring based method for online reliability evaluation of wind turbines. A blade system of a wind turbine is used as an illustrative example to demonstrate the proposed approach. / SURE Grant scheme. SESAME H2020 Project under Grant 101017258.

It is important to manage iterative projects in a way to maximize quality and minimize cost. To achieve high quality, accurate project estimates are of high importance. It is challenging to predict the effort that is required to perform test activities in an iterative development. If testers put extra effort in testing then schedule might be delayed, however, if testers spend less effort then quality could be affected. Currently there is no model for test effort prediction in iterative development to overcome such challenges. This paper introduces and validates a dynamic Bayesian network to predict test effort in iterative software development. In this research work, the proposed framework is evaluated in a number of ways: First, the framework behavior is observed by considering different parameters and performing initial validation. Then secondly, the framework is validated by incorporating data from two industrial projects. The accuracy of the results has been verified through different prediction accuracy measurements and statistical tests. The results from the verification confirmed that the framework has the ability to predict test effort in iterative projects accurately.

Tragically, traffic accidents involving drivers, motorcyclists and pedestrians result in thousands of fatalities worldwide each year. For this reason, making improvements to road safety and saving people's lives is an international priority. In recent years, this aim has been supported by Intelligent Transport Systems, offering safety systems and providing an intelligent driving environment. The development of wireless communications and mobile ad hoc networks has led to improvements in intelligent transportation systems heightening these systems' safety. Vehicular ad hoc Networks comprise an important technology; included within intelligent transportation systems, they use dedicated short-range communications to assist vehicles to communicate with one another, or with those roadside units in range. This form of communication can reduce road accidents and provide a safer driving environment. A major challenge has been to design an ideal system to filter relevant contextual information from the surrounding environment, taking into consideration the contributory factors necessary to predict the likelihood of a crash with different levels of severity. Designing an accurate and effective pre-crash system to avoid front and back crashes or mitigate their severity is the most important goal of intelligent transportation systems, as it can save people's lives. Furthermore, in order to improve crash prediction, context-aware systems can be used to collect and analyse contextual information regarding contributory factors. The crash likelihood in this study is considered to operate within an uncertain context, and is defined according to the dynamic interaction between the driver, the vehicle and the environment, meaning it is affected by contributory factors and develops over time. As a crash likelihood is considered to be an uncertain context and develops over time, any usable technology must overcome this uncertainty in order to accurately predict crashes. This thesis presents a context-aware pre-crash collision prediction system, which captures information from the surrounding environment, the driver and other vehicles on the road. It utilises a Dynamic Bayesian Network as a reasoning model to predict crash likelihood and severity level, whether any crash will be fatal, serious, or slight. This is achieved by combining the above mentioned information and performing probabilistic reasoning over time. The thesis introduces novel context aware on-board unit architecture for crash prediction. The architecture is divided into three phases: the physical, the thinking and the application phase; these which represent the three main subsystems of a context-aware system: sensing, reasoning and acting. In the thinking phase, a novel Dynamic Bayesian Network framework is introduced to predict crash likelihood. The framework is able to perform probabilistic reasoning to predict uncertainty, in order to accurately predict a crash. It divides crash severity levels according to the UK department for transport, into fatal, serious and slight. GeNIe version 2.0 software was used to implement and verify the Dynamic Bayesian Network model. This model has been verified using both syntactical and real data provided by the UK department for transport in order to demonstrate the prediction accuracy of the proposed model and to demonstrate the importance of including a large amount of contextual information in the prediction process. The evaluation of the proposed system delivered high-fidelity results, when predicting crashes and their severity. This was judged by inputting different sensor readings and performing several experiments. The findings of this study has helped to predict the probability of a crash at different severity levels, accounting for factors that may be involved in causing a crash, thereby representing a valuable step towards creating a safer traffic network.

Bayesian networks are a means to study data. A Bayesian network gives structure to data by creating a graphical system to model the data. It then develops probability distributions over these variables. It explores variables in the problem space and examines the probability distributions related to those variables. It conducts statistical inference over those probability distributions to draw meaning from them. They are good means to explore a large set of data efficiently to make inferences. There are a number of real world applications that already exist and are being actively researched. This paper discusses the theory and applications of Bayesian networks.

Ce travail de recherche est consacré à la formalisation et à la réalisation d’un processus de diagnostic pluridisplinaire. La particularité d’un tel diagnostic résulte du fait qu’il nécessite de nombreux spécialistes, chacun ayant des connaissances sur leur domaine. Le problème principal réside dans les interconnexions entre les domaines. Ces interconnexions peuvent ou non être connues et influer sur le diagnostic. Dans ce manuscrit, nous proposons de réaliser un diagnostic pluridisciplinaire l’aide d’un système multi-agents. Les agents élaborent un diagnostic local à un domaine puis, fusionnent leurs diagnostics afin d’obtenir le diagnostic pluridisciplinaire. Dans ce but, nous proposons un cadre d’argumentation et une méthode de fusion des diagnostics. Ensemble, ces deux propositions forment le modèle ANDi. / Sharing opinions among different participants is a useful and common way to build a constructive argumentation in order to solve complex problems that require the confrontation of different discipline areas. In such settings, experts build different arguments in relation to their own discipline area, then share and confront them to the other experts’ opinions. In this report we present an argumentative framework ANDi based on a multi-agent approach and Bayesian networks. In this framework, the agents support the elaboration of a global diagnostic from local ones. Local diagnostics are resulting of argumentations between group of experts from the same discipline area. We illustrate the use of this argumentation framework on the domain of fault diagnosis.