Deciding the optimal media advertisement spending is a complex issue that many companies today are facing. With the rise of new ways to market products, the choices can appear infinite. One methodical way to do this is to use Marketing Mix Modelling (MMM), in which statistical modelling is used to attribute sales to media spendings. However, many problems arise during the modelling. Modelling and mitigation of uncertainty, time-dependencies of sales, incorporation of expert information and interpretation of models are all issues that need to be addressed. This thesis aims to investigate the effectiveness of eight different statistical and machine learning methods in terms of prediction accuracy and certainty, each one addressing one of the previously mentioned issues. It is concluded that while Shapley Value Regression has the highest certainty in terms of coefficient estimation, it sacrifices some prediction accuracy. The overall highest performing model is the Bayesian hierarchical model, achieving both high prediction accuracy and high certainty.

Object of the research. The research object includes a CCA and a model of a web-based decision support system developed by integrating basic knowledge in the sphere of construction management, civil law, mathematical methods and advanced computer technologies; the model system enables effective implementation of aims of various parties of a CCA. Aim and tasks of the work. The main aim of the dissertation is to develop a theoretical model of a web-based decision support system for legal regulation of CCAs, taking in to consideration principles of construction management, improvement of effectiveness of legal regulation of CCAs, employing computer technologies, theoretical assumptions related to development of e-business systems, and etc. In order to make an effective theoretical model of a web-based decision support system for legal regulation of CCAs, the following tasks are stated: • To analyse importance of e-business in improvement of effectiveness of preparation of CCAs. • To make a feasibility study of the potential of e-business principles, of factors determining effectiveness and of decision support measures in preparation of CCAs. • To analyse principles of operating decision support systems intended for construction, law and agreement preparation. • To prepare several models for evaluation of CCAs and to select the best. • To analyse the potential of application of currently used multiple criteria analysis methods in a web-based decision support system for legal... [to full text]

La caractérisation de texture est centrale dans de nombreuses applications liées au traitement d’images. L’analyse de textures peut être envisagée dans le cadre mathématique de l’analyse multifractale qui permet d’étudier les fluctuations de la régularité ponctuelle de l’amplitude d’une image et fournit les outils pratiques pour leur évaluation grâce aux coefficients d’ondelettes ou aux coefficients dominants. Bien que mise à profit dans de nombreuses applications, l’analyse multifractale souffre à présent de deux limitations majeures. Premièrement, l’estimation des paramètres multifractaux reste délicate, notamment pour les images de petites tailles. Deuxièmement, l’analyse multifractale a été jusqu’à présent uniquement considérée pour l’analyse univariée d’images, alors que les données à étudier sont de plus en plus multivariées. L’objectif principal de cette thèse est la mise au point de contributions pratiques permettant de pallier ces limitations. La première limitation est abordée en introduisant un modèle statistique générique pour le logarithme des coefficients dominants, paramétrisé par les paramètres multifractaux d’intérêt. Ce modèle statistique permet de contrebalancer la variabilité résultant de l’analyse d’images de petite taille et de formuler l’estimation dans un cadre bayésien. Cette approche aboutit à des procédures d’estimation robustes et efficaces, que ce soit pour des images de petites ou grandes tailles. Ensuite, l’analyse multifractale d’images multivariées est traitée en généralisant ce cadre bayésien à des modèles hiérarchiques capables de prendre en compte l’hypothèse d’une évolution lente des propriétés multifractales d’images multi-temporelles ou multi-bandes. Ceci est réalisé en définissant des lois a priori reliant les propriétés dynamiques des paramètres multifractaux des différents éléments composant le jeu de données. Différents types de lois a priori sont étudiés dans cette thèse au travers de simulations numériques conduites sur des images multifractales multivariées synthétiques. Ce travail est complété par une étude du potentiel apport de l’analyse multifractale et de la méthodologie bayésienne proposée pour la télédétection à travers l’exemple de l’imagerie hyperspectrale. / Texture characterization is a central element in many image processing applications. Texture analysis can be embedded in the mathematical framework of multifractal analysis, enabling the study of the fluctuations in regularity of image intensity and providing practical tools for their assessment, the coefficients or wavelet leaders. Although successfully applied in various contexts, multi fractal analysis suffers at present from two major limitations. First, the accurate estimation of multifractal parameters for image texture remains a challenge, notably for small sample sizes. Second, multifractal analysis has so far been limited to the analysis of a single image, while the data available in applications are increasingly multivariate. The main goal of this thesis is to develop practical contributions to overcome these limitations. The first limitation is tackled by introducing a generic statistical model for the logarithm of wavelet leaders, parametrized by multifractal parameters of interest. This statistical model enables us to counterbalance the variability induced by small sample sizes and to embed the estimation in a Bayesian framework. This yields robust and accurate estimation procedures, effective both for small and large images. The multifractal analysis of multivariate images is then addressed by generalizing this Bayesian framework to hierarchical models able to account for the assumption that multifractal properties evolve smoothly in the dataset. This is achieved via the design of suitable priors relating the dynamical properties of the multifractal parameters of the different components composing the dataset. Different priors are investigated and compared in this thesis by means of numerical simulations conducted on synthetic multivariate multifractal images. This work is further completed by the investigation of the potential benefit of multifractal analysis and the proposed Bayesian methodology for remote sensing via the example of hyperspectral imaging.

Urbanisation alters the hydrologic response of a catchment, resulting in increased runoff rates and volumes, and loss of infiltration and base flow. Quantification of uncertainties is important in hydrologic designs of urban infrastructure. Major sources of uncertainty in the Intensity Duration Frequency (IDF) relationships are due to insufficient quantity and quality of data leading to parameter uncertainty and, in the case of projections of future IDF relationships under climate change, uncertainty arising from use of multiple General Circulation Models (GCMs) and scenarios. The work presented in the thesis presents methodologies to quantify the uncertainties arising from parameters of the distribution fitted to data and the multiple GCMs using a Bayesian approach. High uncertainties in GEV parameters and return levels are observed at shorter durations for Bangalore City. Twenty six GCMs from the CMIP5 datasets, along with four RCP scenarios are considered for studying the effects of climate change. It is observed that the uncertainty in short duration rainfall return levels is high when compared to the longer durations. Further it is observed that parameter uncertainty is large compared to the model uncertainty. Disaggregation of precipitation extremes from larger time scales to smaller time scales when the extremes are modeled with the GPD is burdened with difficulties arising from varying thresholds for different durations. In this study, the scale invariance theory is used to develop a disaggregation model for precipitation extremes exceeding specified thresholds. A scaling relationship is developed for a range of thresholds obtained from a set of quantiles of non-zero precipitation of different durations. The disaggregation model is applied to precipitation datasets of Berlin City, Germany and Bangalore City, India. From both the applications, it is observed that the uncertainty in the scaling exponent has a considerable effect on uncertainty in scaled parameters and return levels of shorter durations. A Bayesian hierarchical model is used to obtain spatial distribution of return levels of precipitation extremes in urban areas and quantify the associated uncertainty. Applicability of the methodology is demonstrated with data from 19 telemetric rain gauge stations in Bangalore City, India. For this case study, it is inferred that the elevation and mean monsoon precipitation are the predominant covariates for annual maximum precipitation. For the monsoon maximum precipitation, it is observed that the geographic covariates dominate while for the summer maximum precipitation, elevation and mean summer precipitation are the predominant covariates. In this work, variation in the dependence structure of extreme precipitation within an urban area and its surrounding non-urban areas at various durations is studied. The Berlin City, Germany, with surrounding non-urban area is considered to demonstrate the methodology. For this case study, the hourly precipitation shows independence within the city even at small distances, whereas the daily precipitation shows a high degree of dependence. This dependence structure of the daily precipitation gets masked as more and more surrounding non-urban areas are included in the analysis. The geographical covariates are seen to be predominant within the city and the climatological covariates prevail when non-urban areas are added. These results suggest the importance of quantification of dependence structure of spatial precipitation at the sub-daily timescales, as well as the need to more precisely model spatial extremes within the urban areas. The work presented in this thesis thus contributes to quantification of uncertainty in precipitation extremes through developing methodologies for generating probabilistic future IDF relationships under climate change, spatial mapping of probabilistic return levels and modeling dependence structure of extreme precipitation in urban areas at fine resolutions.

Na engenharia, existem diversas técnicas de análise estrutural que conduzem à obtenção de esforços e deslocamentos com considerável grau de eficácia e acurácia e, dentre estas, há a técnica de SUBESTRUTURAÇÃO. Esta técnica consiste na divisão e análise de um modelo estrutural completo, através da divisão de seu domínio em SUBESTRUTURAS, permitindo uma considerável eficiência nos cálculos e na avaliação de modelos estruturais combinados com a análise hierárquica de estruturas. Neste trabalho, seguindo-se a metodologia de pesquisa, que envolveu revisão bibliográfica e desenvolvimento de modelos computacionais, apresentaram-se as principais técnicas de subestruturas - seção 1, definições e formulação teórica dos principais métodos - desenvolvidos por Przemienieck (1963), Rosen e Rubinstein (1968), Rubinstein (1970) e El-Sayed e Hsiung (1990) - e a conceituação da análise hierárquica de estruturas com as respectivas aplicações combinadas com técnica de subestruturação para análise de fissuras em estruturas aeronáuticas, proposta de Starnes e Britt (1991) e aplicada por Potyondy (1993) - seção 2. Nesse contexto, na seção 3 se propôs a metodologia hierárquica para análise de estruturas típicas da construção civil, utilizando técnicas de subestruturação. Apresentaram-se instrumentos úteis à respectiva aplicação, dentre os quais se destaca a proposta de metodologia simplificada de análise hierárquica. Algumas precauções a serem adotadas na aplicação da metodologia também foram apresentadas. Na seção 4, desenvolveram-se cinco estudos de caso aplicando-se as metodologias, instrumentos e precauções desenvolvidos na seção 3, destacando-se a eficácia das metodologias. Conclui-se a pesquisa na seção 5 sendo que, na seção 6, indicam-se novas linhas de pesquisas a serem desenvolvidas sobre a metodologia proposta na pesquisa. / In engineering, there are several structural analysis techniques leading to obtain displacements and efforts with a considerable degree of efficacy and accuracy, and, among these, there is the SUBSTRUCTURING technique. This method consists in the division and analysis of a complete structural model, by subdividing his field in SUBSTRUCTURES, allowing a considerable efficiency in calculations and evaluation of structural models combined with hierarchical structures analysis. In this work, following the methodology research, which involved a literature review and computational models development, the key techniques of substructures were presented - section 1, as well as main methods definitions and theoretical formulation - developed by Przemienieck (1963), Rosen and Rubinstein (1968), Rubinstein (1970) and El-Sayed and Hsiung (1990) - and structural hierarchical analysis concept with their applications combined with aircraft structures cracks substructuring technical analysis, proposed by Starnes and Britt (1991), and implemented by Potyondy (1993) - Section 2. In this context, a civil construction typical structures analysis methodology was proposed in section 3. Besides, many useful tools for their implementation were presented, among which stands out the simplified hierarchy analysis methodology proposal. Some precautions in the methodology applying were also presented. In section 4, five case studies were developed, applying the methodology, tools and precautions explained in section 3, especially the methodologies effectiveness. The research is concluded in section 5, and new research lines, to be undertaken on the methodology proposed, were outpointed in section 6.

O design de sistemas automatizados inteligentes é fortemente dependente da etapa inicial de análise de requisitos, que além de suprimir possíveis incongruências - próprias desta fase inicial - provê um modelo inicial e funcional do sistema a ser implementado, capaz de orientar a definição dos parâmetros (design parameters) e em seguida a própria implementação. Um grande esforço tem sido empregado na área de Inteligência Artificial para definir planejadores automáticos confiáveis que possam ser usados na solução de problemas reais, que geralmente possuem um número elevado de parâmetros. Isto leva a uma situação onde métodos formais, geralmente aplicados em abordagens independentes de domínio, precisam ser aliados a métodos mais pragmáticos para produzir bons resultados. Seguindo esse princípio,o presente trabalho propõe um método de projeto no qual o usuário podem fazer a aquisição de conhecimento, modelar o domínio (tanto o domínio de aplicação quanto o domínio do problema de planejamento), fazer uma análise dinâmica do modelo e eventualmente verificá-lo usando linguagens conhecidas como UML, as Redes de Petri, e HTN, mantendo as características hierárquicas do problema. Esta abordagem demanda novos planejadores automáticos que consideram a abstração do sistema hierárquico, que é derivado de um modelo hierárquico de requisitos e de uma análise unificada feita também em redes de Petri hierárquicas. / The design of intelligent systems is strongly dependent of the requirements analysis initial stage, wich can remove possible inconsistencies - own this early stage - providing a primary functional model of the system being implemented. Besides that, it can be able to guide the definition of the parameters (design parameters) and then the implementation itself. A lot of effort has been employed in Artificial Intelligence field to define reliable automated planning systems that can be used to solve real problems, which generally have a large number of parameters. This leads to a situation where formal methods, generally applied to solve domain-independent problems, need to be combined with more pragmatic methods to produce good results. Following this principle, this work proposes a design method in which the user can make the acquisition of knowledge, model the domain (both the work domain and the planning problem domain), make a dynamic analysis of the model and eventually simulate it using well-kown language as UML, Petri Nets, and HTN, maintaining the hierarchical characteristics of the problem. This approach demands new automated planners who consider the abstraction of hierarchical system, which is derived from a hierarchical model requirements and an unified analysis made in hierarchical Petri nets.

Conjoint analysis is a popular method in consumer preferences research. One of the factors that caused the increasing popularity of this method in recent years is wide use of hierarchical Bayesian models which has been found invaluable in solving the problem of how to obtain reliable estimates of individual preferences without need for overloading respondents with too many conjoint tasks. First goal of my dissertation was to confirm whether the use of Bayesian models is the best choice under all circumstances or whether there are some limitations of this approach. For this purpose I conducted a study based on simulated datasets. Algorithm used enabled generation of datasets that differed in several parameters of interest but which were most comparable in other aspects. Results show that hierarchical models represent choice leading to highest accuracy in predicting respondents' choices in holdout tasks. Use of hierarchical models is most beneficial in the situation of strongly heterogeneous population yet limited amount of available data. In these cases we are able to capture the structure of heterogeneity with significantly lower number of choice task necessary from each respondent. Second goal of the dissertation was to answer the question whether we can increase also the effectiveness of the questioning in conjoint analysis by adding several direct questions. Suggested hybrid choice-based conjoint method (HCBC) combines conjoint analysis tasks with direct questions regarding the preference of levels for each attribute. These are used during the estimation of the model and for increasing the effectiveness if the conjoint analysis tasks design. The HCBC was compared with traditional choice-based conjoint (CBC) and adaptive choice-based conjoint (ACBC) based on practical study involving 421 respondents randomly assigned in one of three test groups. Suggested method has been found as useful alternative that can help with reducing number of choice task needed and as a solution for some situations when diverse importance of the attributes tested does not allow for indirect estimation of preferences with respect to all attributes tested.

Submitted by Renata Lopes (renatasil82@gmail.com) on 2015-12-10T11:17:15Z No. of bitstreams: 1 luisantoniofajardopontes.pdf: 1987237 bytes, checksum: a9022a27e3eff6caa86aab0734002199 (MD5) / Rejected by Adriana Oliveira (adriana.oliveira@ufjf.edu.br), reason: Renata, favor corrigir: o caso do Índice de Desenvolvimento da Educação Básica, o Ideb, para: IDEB on 2015-12-10T14:07:41Z (GMT) / Submitted by Renata Lopes (renatasil82@gmail.com) on 2015-12-10T14:13:15Z No. of bitstreams: 1 luisantoniofajardopontes.pdf: 1987237 bytes, checksum: a9022a27e3eff6caa86aab0734002199 (MD5) / Approved for entry into archive by Adriana Oliveira (adriana.oliveira@ufjf.edu.br) on 2015-12-10T14:13:58Z (GMT) No. of bitstreams: 1 luisantoniofajardopontes.pdf: 1987237 bytes, checksum: a9022a27e3eff6caa86aab0734002199 (MD5) / Made available in DSpace on 2015-12-10T14:13:58Z (GMT). No. of bitstreams: 1 luisantoniofajardopontes.pdf: 1987237 bytes, checksum: a9022a27e3eff6caa86aab0734002199 (MD5) Previous issue date: 2015-05-15 / O presente estudo tem por objetivo analisar alguns mecanismos específicos de aferição da qualidade educacional que usualmente vêm sendo propostos por diversos sistemas educacionais no Brasil e, mais particularmente, nas escolas públicas de Minas Gerais, bem como propor alternativas de seu aperfeiçoamento. Neste sentido, discorre-se primeiramente sobre a questão da eficácia escolar e também sobre a maneira como este tópico vem ganhando uma importância cada vez maior na agenda das reformas educacionais tanto fora quanto dentro do Brasil. A seguir, apresenta-se o caso do Índice Brasileiro de Desenvolvimento da Educação Básica, o Ideb, discorrendo-se sobre a sua metodologia de elaboração e também sobre sua interpretação. Posteriormente, apresentam-se conclusões, específicas para as escolas públicas de Minas Gerais, acerca dos efeitos indesejáveis sobre as metas do Ideb causados, em parte, por uma certa instabilidade dos resultados educacionais, que fazem com que, a cada edição dessa avaliação transversal, ocorram desvios ou flutuações dos resultados em relação a uma tendência de crescimento prevista por um modelo longitudinal hierárquico. Utilizando-se então procedimentos associados a esta mesma metodologia, é proposto um modelo inédito de determinação das metas do Ideb, por meio de uma correção periódica que se faz sobre elas, com base nos resultados atualizados das avaliações, parte delas relacionadas, no caso deste exemplo de Minas Gerais, aos resultados anuais dos testes do Sistema Mineiro de Avaliação da Educação Básica, o SIMAVE, na 8ª série ou 9º ano do Ensino Fundamental. E, finalmente, estudam-se também casos específicos de melhoria significativa da qualidade das decisões sobre o desempenho educacional dos estabelecimentos assim avaliados, com base neste novo modelo. / This study aims to analyze some specific mechanisms of educational quality measurement which have been proposed by several education systems in Brazil and more specifically in Minas Gerais State public schools, as well as propose some alternatives for their improvement. In this sense, some questions about school effectiveness are proposed, along with some considerations about the way how this topic has acquired an increasing relevance in educational reform agendas both inside and outside Brazil. The specific case of the Brazilian Index of Basic Education Development – originally the Índice Brasileiro de Desenvolvimento da Educação Básica, or Ideb – is also presented, together with some considerations about its development methodology and interpretation. Later some specific conclusions about the Ideb and Minas Gerais State public schools are also discussed, related to the undesirable effect that has been detected on their respective Ideb targets, partly caused by a certain instability of educational results, by which it is possible to detect and gauge some deviations or fluctuations of the results in relation to a growth trend model at each year of the historical evaluation series, according to a longitudinal hierarchical model. The latter methodology is also employed in order to propose a new model of Ideb target setting, which is accomplished by their respective periodic correction based on updated results obtained by the same schools in other state standardized tests, such as the case, in Minas Gerais, of those related to the Sistema Mineiro de Avaliação da Educação Básica – or Minas Gerais Basic Education Evaluation Sistem, the SIMAVE –, for the 9th year or 8th grade of Fundamental Level. Finally, the text presents and discusses some specific cases of occurrence of a significant quality improvement related to decisions that can be made about the educational performance of the same educational institutions evaluated according to this model.

Les modèles de croissance de plantes peuvent être utilisés afin de prédire des quantités d’intérêt ou évaluer la variabilité génotypique au sein d’une population de plantes ; ce double usage est mis en évidence au sein de ce travail. Trois modèles de plantes sont ainsi considérés (LNAS pour la betterave et le blé, GreenLab pour Arabidopsis thaliana) au sein du cadre mathématique des modèles à espace d’états généraux.Une nouvelle plate-forme de calcul générique pour la modélisation et l’inférence statistique (ADJUSTIN’) a été développée en Julia, permettant la simulation des modèles de croissance de plantes considérés ainsi que l’utilisation de techniques d’estimation de pointe telles que les méthodes de Monte Carlo par chaînes de Markov ou de Monte Carlo séquentielles.L’inférence statistique au sein des modèles de croissance de plantes étant de première importance pour des applications concrètes telles que la prédiction de rendement, les méthodes d’estimation de paramètres et d’états au sein de modèles à espaces d’états et dans un cadre bayésien furent tout d’abord étudiées, et plusieurs cas d’étude pour les plantes considérées sont analysés pour le cas d’une plante individuelle.La caractérisation de la variabilité au sein d’une population de plantes est envisagée à travers les distributions des paramètres de population au sein de modèles hiérarchiques bayésiens. Cette approche requérant l’acquisition de nombreuses données pour chaque individu, un algorithme de segmentation-suivi pour l’analyse d’images d’Arabidopsis thaliana, obtenues grâce au Phénoscope, une plate-forme de phénotypage à haut rendement de l’INRA Versailles, est proposé.Finalement, l’intérêt de l’utilisation des modèles hiérarchiques bayésiens pour la mise en évidence de la variabilité au sein d’une population de plantes est discutée. D’abord par l’étude de différents scénarios sur des données simulées, et enfin en utilisant les données expérimentales obtenues à partir de l’analyse d’images pour une population d’Arabidopsis thaliana comprenant 48 individus. / Plant growth models can be used in order to predict quantities of interest or assess the genotypic variability of a population of plants; this dual use is emphasized throughout this work.Three plant growth models are therefore considered (LNAS for sugar beet and wheat, GreenLab for Arabidopsis thaliana) within the mathematical framework of general state space models.A new generic computing platform for modelling and statistical inference (ADJUSTIN’) has been developed in Julia, allowing to simulate the plant growth models considered as well as the use of state-of-the-art estimation techniques such as Markov chain Monte Carlo and sequential Monte Carlo methods.Statistical inference within plant growth models is of primary importance for concrete applications such as yield prediction, parameter and state estimation methods within general state-space models in a Bayesian framework were first studied and several case studies for the plants considered are then investigated in the case of an individual plant.The characterization of the variability of a population of plants is envisioned through the distributions of parameters using Bayesian hierarchical models. This approach requiring the acquisition of numerous data for each individual, a segmentation-tracking algorithm for the analysis of images of Arabidopsis thaliana, obtained thanks to the Phenoscope, a high-throughput phenotyping platform of INRA Versailles, is proposed.Finally, the interest of using Bayesian hierarchical models to evidence the variability of a population of plants is discussed. First through the study of different scenarios on simulated data, and then by using the experimental data acquired via image analysis for the population of Arabidopsis thaliana comprising 48 individuals.

Thesis ((Ph.D. (Mathematics Education)) -- University of Limpopo, 2016 / The focus in this thesis was on the approaches that seek to compare two study arms in the absence of randomisation when the interclass correlation coefficient is greater than zero. Many reports on performance of learners in Grade 12 have used ordinary regression models (such as logistic regression model and linear regression models) which ignore clustering effect, and descriptive statistics (e.g., averages and standard deviations for continuous variables, and proportions expressed as percentages and frequencies). These models do not only bias point estimates but also give falsely narrow confidence intervals. The study was applied to two of the nine provinces of South Africa: Gauteng Province and Western Cape Province in 2008, 2009 and 2010 academic years. Causal models, and in particular, hierarchical models (or disaggregated approach), unlike descriptive analyses, are more powerful as they are able to adjust for individual covariates. For the analysis of continuous variables; Western Cape Province was expected to significantly score higher marks than Gauteng Province in 2008 (Crude estimate: 0.782) and 2009 (Crude estimate: 0.957 ) while Gauteng Province was expected to score higher marks than Western Cape Province in 2010 (Crude estimate: −0.302). Adjusted models indicate that Western Cape Province performed better than Gauteng Province in 2008 and 2009 but not in 2010 where Gauteng Province performed better than Western Cape Province after adjusting for gender. In case of binary outcome; the crude estimates favoured Western Cape Province than Gauteng Province in 2008 (Odds ratio = 1.16) and 2009 (Odds ratio = 1.19). Otherwise, the crude estimates favoured Gauteng Province in 2010 (Odds ratio = 0.11). The proportion of female learners in Gauteng Province ranged between 54.48% and 54.99%, while in Western Cape Province it ranged between 56.78% and 57.16%, in 2008 through 2010 academic years. Proportion of female learners in Western Cape Province were found to be higher than those in Gauteng during this period. At least 70.42% of learners in Gauteng and at least 73.96% of learners in Western Cape Province passed Grade 12 during the years 2008 to 2010. Through the application of causal model we have learned that although gender is not a significant predictor of the overall learner performance in Grade 12, the effect of gender gave the mixed findings depending on the nature of the outcome. The xi effect of gender on continuous endpoint (marks) suggests that a model of single-sex classrooms or single-sex schools may be adopted so as to mitigate the inherent perceptions and stereotype regarding learner-gender. However, the results based on binary endpoint (pass/not pass) suggest that coeducation system is the best bet. A school quintile is a significant predictor of the overall learner performance in the two provinces. The resourceful schools are more likely to produce learners with higher marks. Also, the resourced schools than the less or under resourced schools are more likely to produce the favourable results (higher marks (%) or/and pass) in the two provinces.