No description available.

Managers in organizations make investment decisions all the time. These decisions have an impact on the bottom-line profits and on the market penetration of the organization. Some decisions have more impact than others do and not all such decisions are evaluated for their impact. The Service-Profit Chain (SPC) framework brings together several components like operational attributes, customer perceptions, customer behavioral intentions and customer loyalty to evaluate the service operation. This research augments the SPC with another component — uncontrollable factors (environmental variables and competition) that are exogenous to the operation but definitely have an effect on the service delivery process. Further, this research develops a dynamic model to evaluate investments made in operational attributes (e.g. number of tellers in a bank, number of airline flight options to a particular city available to customers) and determine the behavior of customer perceptions, customer intentions, customer loyalty, profit, market penetration and marginal rate of return over time. The above is accomplished by incorporating a hill-climbing algorithm into the dynamic SPC model. This hill-climbing algorithm senses the current state of the system and compares it to a certain goal to determine the discrepancies and make additional interventions. The objective is to determine an optimal path to steady state and to evaluate if certain goals are realistic. Next, the Service Sustainability Chain is developed to be applicable to training services. This is accomplished by building key relationships specific to training services into separate modules. The Dynamic SPC module is based on the SPC framework. The Customer Base Growth module captures the structure for referrals and how this enables the growth of the customer base mimicking the infectious model for epidemic diseases in the literature. A methodology based on Chi-Square Automatic Interaction Detection (CHAID) and Structural Equation Modeling (SEM) developed to explore, uncover and identify relationships and mathematical equations is used to determine the structural input-output representation of the SPC. Next, the model and the methodology developed are applied to a case study in a training services organization, simulated and validated. / Ph. D.

The aim of this thesis is comparison of selected classification methods which are logistic regression (binary and multinominal), multilayer perceptron and classification trees, CHAID and CRT. The first part is reminiscent of the theoretical basis of these methods and explains the nature of parameters of the models. The next section applies the above classification methods to the six data sets and then compares the outputs of these methods. Particular emphasis is placed on the discriminatory power rating models, which a separate chapter is devoted to. Rating discriminatory power of the model is based on the overall accuracy, F-measure and size of the area under the ROC curve. The benefit of this work is not only a comparison of selected classification methods based on statistical models evaluating discriminatory power, but also an overview of the strengths and weaknesses of each method.

Atualmente existem muitas fontes de dados oriundos de movimentação de pessoas e veículos. Deduzir o comportamento do tráfego a partir desses dados pode ser de grande utilidade para a gestão do tráfego. Entretanto, os métodos existentes podem não abordar todas as dimensões referentes ao comportamento do tráfego, em especial sob efeito de adversidades, como por exemplo - eventos climáticos. O objetivo dessa dissertação é apresentar uma combinação de métodos estatísticos e de análise espacial, para uso na compreensão dos padrões temporais e espaciais de variabilidade das velocidades, em condições normais e sob o impacto da precipitação. O método consiste na mineração de dados de micro trechos de deslocamento (microviagens), através da combinação de estatísticas descritivas, árvores de regressão CHAID (Chi square automatic interaction detection) e análises espaciais (Krigagem). Para a validação do método foram utilizados dados obtidos, durante março de 2014, de operação da frota de ônibus municipal da cidade de São Paulo, fornecidos pela São Paulo Transportes (SPTrans). Os dados de precipitação de chuva foram fornecidos pelo Instituto Nacional de Meteorologia (INMET). Os resultados demonstraram o impacto da precipitação sobre a variabilidade das velocidades por dia da semana e hora do dia. Adicionalmente, com base nos conjuntos de dados classificados via CHAID, foram produzidas representações da distribuição das velocidades sobre a superfície através uso da Krigagem. Observou-se uma consistente queda das velocidades no período vespertino, principalmente entre 16:00 e 19:00 horas, frequentemente agravados pela presença de precipitação. Foram selecionados casos de alta variabilidade, a partir da técnica CHAID, que foram analisados espacialmente. Os efeitos da queda da variabilidade foram observados geograficamente, demonstrando que a infraestrutura de transporte é um importante fator em sua caracterização. O estudo demonstra a viabilidade desta abordagem e a importância de incorporar a geografia no entendimento da variabilidade das velocidades em geral. / Nowadays there is a wide number of movement data sources which can be useful in transportation management. However, the existing methods may not cover all traffic dimensions, especially under inclement weather. The objective of this work is presenting a combination of statistical and spatial methods which allows to understand space and time speed patterns, considering regular operation conditions in comparison with rainfall events. The method consists in sequential bus tracking data segments (micro travel) mining by the combination of descriptive statistics, CHAID (Chi square automatic interaction detection) regression trees and spatial analysis (Kriging). To validate the method, São Paulo bus fleet tracking data provided by SPTrans, and rainfall precipitation, provided by INMET was combined and processed. The results presented rainfall as a relevant factor of speed variability by day of week and time of the day. A consistent speed decrease in the afternoon period, mainly between 4:00 PM and 7:00 PM, was observed. Such effects are more intense in face of rainfall precipitation. The highest variability cases obtained by CHAID classification was selected and spatially analyzed showing the transportation infrastructure is an important variability component. Such results show the importance of a geographic approach over speed variability.

Atualmente existem muitas fontes de dados oriundos de movimentação de pessoas e veículos. Deduzir o comportamento do tráfego a partir desses dados pode ser de grande utilidade para a gestão do tráfego. Entretanto, os métodos existentes podem não abordar todas as dimensões referentes ao comportamento do tráfego, em especial sob efeito de adversidades, como por exemplo - eventos climáticos. O objetivo dessa dissertação é apresentar uma combinação de métodos estatísticos e de análise espacial, para uso na compreensão dos padrões temporais e espaciais de variabilidade das velocidades, em condições normais e sob o impacto da precipitação. O método consiste na mineração de dados de micro trechos de deslocamento (microviagens), através da combinação de estatísticas descritivas, árvores de regressão CHAID (Chi square automatic interaction detection) e análises espaciais (Krigagem). Para a validação do método foram utilizados dados obtidos, durante março de 2014, de operação da frota de ônibus municipal da cidade de São Paulo, fornecidos pela São Paulo Transportes (SPTrans). Os dados de precipitação de chuva foram fornecidos pelo Instituto Nacional de Meteorologia (INMET). Os resultados demonstraram o impacto da precipitação sobre a variabilidade das velocidades por dia da semana e hora do dia. Adicionalmente, com base nos conjuntos de dados classificados via CHAID, foram produzidas representações da distribuição das velocidades sobre a superfície através uso da Krigagem. Observou-se uma consistente queda das velocidades no período vespertino, principalmente entre 16:00 e 19:00 horas, frequentemente agravados pela presença de precipitação. Foram selecionados casos de alta variabilidade, a partir da técnica CHAID, que foram analisados espacialmente. Os efeitos da queda da variabilidade foram observados geograficamente, demonstrando que a infraestrutura de transporte é um importante fator em sua caracterização. O estudo demonstra a viabilidade desta abordagem e a importância de incorporar a geografia no entendimento da variabilidade das velocidades em geral. / Nowadays there is a wide number of movement data sources which can be useful in transportation management. However, the existing methods may not cover all traffic dimensions, especially under inclement weather. The objective of this work is presenting a combination of statistical and spatial methods which allows to understand space and time speed patterns, considering regular operation conditions in comparison with rainfall events. The method consists in sequential bus tracking data segments (micro travel) mining by the combination of descriptive statistics, CHAID (Chi square automatic interaction detection) regression trees and spatial analysis (Kriging). To validate the method, São Paulo bus fleet tracking data provided by SPTrans, and rainfall precipitation, provided by INMET was combined and processed. The results presented rainfall as a relevant factor of speed variability by day of week and time of the day. A consistent speed decrease in the afternoon period, mainly between 4:00 PM and 7:00 PM, was observed. Such effects are more intense in face of rainfall precipitation. The highest variability cases obtained by CHAID classification was selected and spatially analyzed showing the transportation infrastructure is an important variability component. Such results show the importance of a geographic approach over speed variability.

Wire Electrical Discharge Machining (WEDM) is a nontraditional machining process used for machining intricate shapes in high strength and temperature resistive (HSTR) materials. WEDM provides high accuracy, repeatability, and a better surface finish; however the tradeoff is a very slow machining rate. Due to the slow machining rate in WEDM, machining tasks take many hours depending on the complexity of the job. Because of this, users of WEDM try to predict machining rate beforehand so that input parameter values can be pre-programmed to achieve automated machining. However, partial success with traditional methodologies such as thermal modeling, artificial neural networks, mathematical, statistical, and empirical models left this problem still open for further research and exploration of alternative methods. Also, earlier efforts in applying the decision tree rule induction algorithms for predicting the machining rate in WEDM had limitations such as use of coarse grained method of discretizing the target and exploration of only C4.5 as the learning algorithm. The goal of this dissertation was to address the limitations reported in literature in using decision tree rule induction algorithms for WEDM. In this study, the three decision tree inductive algorithms C5.0, CART and CHAID have been applied for predicting material removal rate when the target was discretized into varied number of classes (two, three, four, and five classes) by three discretization methods. There were a total of 36 distinct combinations when learning algorithms, discretization methods, and number of classes in the target are combined. All of these 36 models have been developed and evaluated based on the prediction accuracy. From this research, a total of 21 models found to be suitable for WEDM that have prediction accuracy ranging from 71.43% through 100%. The models indentified in the current study not only achieved better prediction accuracy compared to previous studies, but also allows the users to have much better control over WEDM than what was previously possible. Application of inductive learning and development of suitable predictive models for WEDM by incorporating varied number of classes in the target, different learning algorithms, and different discretization methods have been the major contribution of this research.

Made available in DSpace on 2016-04-28T20:56:28Z (GMT). No. of bitstreams: 1 Fabio Alexandre Ferreira Gusmao.pdf: 1672138 bytes, checksum: 11bc28604f163a6e0886cb243d132e2b (MD5) Previous issue date: 2011-10-10 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / Brazil has reached the 21st century providing universal access to elementary education to almost the entire population of the 7 to 14 age group. The 21st century was unquestionably a period of major economic advances, increase in the population s political participation, and consolidating democracy. However, many educational indicators are still dismal, such as the illiteracy rate for the population over 25 years, the student turnover rate, and access rates of students who finish elementary and middle school. The goals of this study are: 1) to establish the relationship between the characteristics of students and schools and the performance results of 8th grade students in Science, and 2) to show how social inequalities are mirrored in the educational system as found by Bourdieu and Passeron (2009). This study used data from the 1999 SAEB (the Brazilian Elementary Education Assessment System) on students, teachers, principal, and school facilities. The data were used to understand the relation and/or correlation between independent variables, indicators of economic and social inequality (GDP, HDI, Gini coefficient) with the dependent variable. This analysis enabled us to capture the impact of different independent variables on student performance in the SAEB database. Parametric and nonparametric tests were subsequently applied to verify whether the differences found were significant. CHAID method was employed to find the Best predictors of performance in Science. In regard to the level of the students, the results show that all the differences found among independent variables and the average proficiency rate in Science were statistically significant. The CHAID analysis highlighted that the following independent variables school system, income, cultural level, flunking rate, and doing homework are statistically related to the average Science proficiency rate. Concerning the level of the schools, results show that school organization and management, school resources, teacher education and pay, and school environment are statistically related to the average Science proficiency rate. In terms of the level of the states, results show that the correlation among social and economic inequality indicators with performance in Science is influenced by the economic, social, and educational situation of Brazilian states and region. CHAID method proved to be effective in possible crossings of data, thus rejecting non-significant variation in data crossing, and focusing its subdivisions on variations resulting from data crossings potentially significant to this study and on identifying predictors of Science learning based on 1999 SAEB data / O Brasil chega ao século XXI praticamente com a universalização do acesso ao ensino fundamental da população da faixa etária dos 7 aos 14 anos. Sem dúvida, um século que se inicia com grandes avanços na área econômica, na participação política da população e na consolidação da democracia; mas que ainda apresenta indicadores educacionais sofríveis, como o índice de analfabetismo da população com mais de 25 anos, a movimentação escolar e as taxas de acesso dos concluintes do ensino fundamental e médio. Os objetivos da presente pesquisa foram: 1º) estabelecer relações entre as características dos alunos e das escolas aos resultados do desempenho de Ciências dos alunos da 8ª série do ensino fundamental, e 2º) mostrar como as desigualdades sociais são reproduzidas no sistema educacional como afirmado por Bourdieu e Passeron (2009). Neste estudo foi utilizado os dados do SAEB de 1999 referentes aos dados coletados de alunos, professores, diretores e instalações físicas das escolas. Os dados foram utilizados para se compreender qual relação e/ou correlação entre as variáveis independentes, os indicadores de desigualdade econômica e social (PIB, IDH, índice de Gini) com a variável dependente. Este procedimento de análise permitiu captar o impacto de diferentes variáveis independentes sobre o desempenho dos alunos a partir da base de dados do SAEB. Foram aplicados testes estatísticos paramétricos e não paramétricos para verificar se as diferenças encontradas foram significativas. Foi utilizado o método CHAID para encontrar os melhores preditores para o desempenho em Ciências. Em relação ao nível dos alunos os resultados mostram que todas as diferenças encontradas entre as variáveis independentes e a proficiência média em Ciências foram estatisticamente significativas. A análise pelo método CHAID destaca que as variáveis independentes rede de ensino, capital econômico e cultural, repetência e fazer dever de casa se relacionam estatisticamente com a proficiência média em Ciências. Em relação ao nível das escolas os resultados indicam que as categorias organização e gestão da escola, recursos escolares, formação e salário docente e clima acadêmico se relacionam estatisticamente com a proficiência média em Ciências. Em relação ao nível dos estados os resultados demonstram que a correlação entre os indicadores de desigualdade econômica e social com o desempenho em Ciências sofre a influência da condição econômica, social e educacional dos estados e regiões brasileiras. O CHAID demonstrou ser eficiente nos cruzamentos possíveis dos dados rejeitando cruzamentos não significativos, e concentrando suas subdivisões em cruzamentos potencialmente significativos para a pesquisa e para identificação de preditores para a aprendizagem em Ciências a partir dos dados do SAEB de 1999

全球化時代與資訊化社會的來臨，電腦與網際網路成為生活中不可或缺的要素，儘管至2008年為止，我國有將近七成的民眾透過網路科技享受到更多的便利性，但社會上仍存在著數位落差（digital divide）的問題，數位落差除了使得資訊窮人（information-poor）不易取得資訊，亦將對其經濟、人權等各方面造成影響。故研究目的在利用資料採礦的應用，配合SPSS Clementine 12.0的軟體，探討數位落差的現象，並嘗試找出形成數位落差的影響原因。 　　本研究主要投入人口統計變數以及生活型態變數，並藉由C5.0決策樹、C&RT分類樹，以及CHAID分類樹建立模型，透過這三個分類迴歸樹的模型，發現到「年齡」、「教育程度」、「地理區域」、「個人資產狀況」、「經濟主要來源：子女」、「個人每月可支配所得」以及「收入來源：薪資」共七項變數同時對民眾是否成為數位落差中的資訊富人（information-rich）有著較重要的影響性，因此，研究最後依據此七項進行政策建議，以提供相關單位之參考。 / In this globalized and informational society, computers and internet networks are essential elements in our daily lives. Until the year 2008, almost 70% of population in Taiwan has enjoyed greater conveniences through networking technologies. However, the issue of “digital divide” remains, where information-poor cannot obtain information easily, and the issue affects the society in terms of economies and human rights. Consequently, the purpose of this research is aimed to find the reasons behind “digital divide” using data-mining techniques with SPSS Clementine 12.0 statistical software. 　　The research will input demographic variables and life-style variables. Using C5.0 decision tree, C&RT tree, and CHAID methodologies to build model, and subsequently discovers that whether the 7 variables - “age”, “level of education”, “location”, “personal asset status”, “main source of income: children”, “monthly personal disposal income” and “source of income: salary” will have significant impacts on information-rich population within “digital divide”. Therefore, the research recommendations will be provided according to the results from these 7 variables.

Made available in DSpace on 2016-04-28T20:56:51Z (GMT). No. of bitstreams: 1 Natalia de Mesquita Matheus.pdf: 3517136 bytes, checksum: 5518b3ee2f114334fb6f19567338c20c (MD5) Previous issue date: 2015-03-19 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / The present study aimed at evaluating the functionality of CHAID to analyze educational data by demonstrating Ideb s sensitivity to different aspects of educational system. The dependent variables were Ideb 2011 from initial and final series of the elementary years; the independent variables were considered separately in a multilevel analysis. At the first level, considered independent variable were the cities former performance measured by the Ideb at 2005, 2007 and 2009; at the second level, the cities budget with education, such as student cost and the percentage of investments at basic education, and others came into focus; and, at the third level, the independent variables considered were demographic data such as region, State, index of human development and influence region. The results shown at the first level indicate a direct relationship between higher scores at the dependent variable and at former performance. Most of the cities with higher scores in 2009 showed higher scores at the Ideb in 2011 and those cities with lower scores in former performances (in 2009, 2007 and, some cases even in 2005) showed lower scores in 2011. At the second level, the results indicate that the effects produced by most recent expenditure (from 2011) are less relevant than those expended earlier (in 2009). Those data suggest that the effects from financial investments in education are / will probably be seen at medium or long term. At the third level, the State was a more significant variable than region and the other tested variables. Statistical algorithms CHAID and exhaustive-CHAID showed to be effective to identify significant variables related to academic performance measured by Ideb, with the production of comprehensive information without losing the specifies of each reality. However, exhaustive-CHAID, turned out to be a more rigorous than CHAID in the selection of variables. It was demonstrated that the multilevel analysis produced different results whern compared with the comprehensive analysis. It suggests an important effects from the interaction among all tested variables. The availability of educational data for identification and analysis of relationships between political actions and its educational outcomes is discussed. Finally,, the contributions of behavior analysis to the interpretation and analysis of people involved at educational policy behavior are considered / O presente trabalho teve como objetivo avaliar a funcionalidade do modelo CHAID para a análise de dados educacionais por meio da demonstração da sensibilidade do Ideb a diferentes aspectos do sistema educacional. As variáveis dependentes do presente estudo foram o Ideb 2011 municipal dos anos iniciais e finais do ensino fundamental e as variáveis independentes foram consideradas separadamente em uma análise multinível. No primeiro nível, considerou-se o desempenho anterior do município no Ideb em 2005, 2007 e 2009; no segundo nível, a execução orçamentária do município, tais como gasto-aluno e porcentagem de investimentos na educação básica, dentre outros; e, no terceiro nível, dados demográficos, como região do país, UF, índice de desenvolvimento humano do município e região de influência. Os resultados no primeiro nível apontam uma relação direta entre índices elevados na VD e em desempenhos anteriores: a maior parte dos municípios com desempenho superior em 2009 apresentaram índices elevados no Ideb em 2011 e aqueles com desempenho inferior em avaliações anteriores (em 2009, 2007 e, em alguns casos, em 2005) apresentaram índices baixos em 2011. No segundo nível, os resultados indicaram que os efeitos gerados por gastos mais recentes (de 2011) são menos relevantes do que aqueles realizados anteriormente (em 2009). Esses dados sugerem que o efeito de investimentos financeiros na educação é/será mais provavelmente observado a médio ou longo prazo. No terceiro nível, a Unidade da Federação se mostrou mais relevante do que a região do país e as demais variáveis testadas. O modelo CHAID, nas duas formas testadas (CHAID e exhaustive-CHAID), se mostrou efetivo para a identificação das variáveis significativamente relacionadas ao desempenho acadêmico, aferido pelo Ideb, produzindo informações abrangentes, mas sem que se percam as especificidades de cada realidade, tendo sido o modelo exhaustive-CHAID mais rigoroso na seleção de variáveis. Demonstrou-se que a análise multinível produz resultados diferentes da análise geral das variáveis, sugerindo importantes efeitos da interação entre elas. Discute-se a disponibilidade de dados educacionais para a identificação e análise de relações entre ações políticas e seus efeitos educacionais. Por fim, nas considerações finais, apresentam-se algumas das contribuições da Análise do Comportamento à análise e interpretação dos comportamentos das pessoas envolvidas na Política Educacional

Abstracts in English, Afrikaans and Northern Sotho / The tax compliance costs of individual taxpayers in South Africa are unknown, and it is essential to determine, amongst other things, whether these costs pose a risk of causing non-compliance, which could have a negative effect on the collection of tax revenue. Hence, the aim of the study was to assess the tax compliance costs of individual taxpayers in South Africa. The assessment of tax compliance costs entailed calculating these costs in relation to the submission of income tax returns and post-filing activities, ascertaining the determinants of these costs and suggesting ways to reduce them. Using data collected from an online survey conducted among 10 260 taxpayers, it was estimated (applying various methods) that income tax compliance costs of individual taxpayers for the 2018 year of assessment were between 3.61% and 5.31% of the personal income tax revenue. These results compared well with ratios reported in most studies conducted in other countries and showed a reduction from the results obtained from 752 taxpayers for the 2017 year of assessment. Chi-square automatic interaction detection (CHAID), a decision tree modelling technique, was used to ascertain the determinants of tax compliance costs and to identify specific groups of taxpayers associated with distinct ranges of the determinants that were statistically significant predictors of tax compliance costs. This breakdown enabled a better understanding of the influence that the specific values of the continuous determinants, such as the service quality rating of the South African Revenue Service (SARS), and the categorical determinants, such as education level and employment status, have on tax compliance costs. The CHAID analysis therefore provided an additional level of insight not possible with regression analysis, enhancing the usability of the results. Employment status and income tax bracket had the strongest association with tax compliance costs (on average, self-employed taxpayers and taxpayers in the highest income tax bracket had the highest total tax compliance costs). Various other determinants, for example, type of assistance obtained, gender, education level, complexity of tax legislation, complexity of SARS guides, and SARS’s service quality rating were also identified in the analyses. / Die belastingvoldoeningskoste van individuele belastingpligtiges in Suid-Afrika is onbekend, en dit is noodsaaklik om, onder andere, te bepaal of hierdie koste ʼn risiko van nievoldoening inhou wat ʼn negatiewe uitwerking op die insameling van belastinginkomste kan hê. Die doelwit van die studie was dus om die belastingvoldoeningskoste van individuele belastingpligtiges in Suid-Afrika te assesseer. Die assessering van belastingvoldoeningskoste behels die berekening van hierdie koste ten opsigte van die indiening van inkomstebelastingopgawes en ná-indieningsaktiwiteite, bepaling van die determinante van hierdie koste en voorstelle van hoe om dit te verminder. Aan die hand van data wat ingesamel is deur ʼn aanlyn opname onder 10 260 belastingpligtiges, is (deur middel van verskeie metodes) geraam dat inkomstebelasting-voldoeningskoste van individuele belastingpligtiges vir die 2018-assesseringsjaar tussen 3.61% en 5.31% van die persoonlike-inkomstebelastinginkomste was. Hierdie resultate het goed vergelyk met verhoudings soos berig in die meeste studies wat in ander lande uitgevoer is en het ʼn afname gewys in die resultate wat by 752 belastingpligtiges vir die 2017-assesseringsjaar bekom is. Chi-kwadraat outomatiese wisselwerkingbespeuring (chi-square automatic interaction detection – CHAID), ʼn besluitnemingskema-modelleringtegniek, is gebruik om die determinante van belastingvoldoeningskoste te bepaal en om spesifieke groepe belastingpligtiges te identifiseer wat statisties-beduidende voorspellers van belastingvoldoeningskoste is. Hierdie uiteensetting het ʼn beter begrip daargestel van die invloed wat die spesifieke waardes op die deurlopende determinante, soos die diensgehaltegradering van die Suid-Afrikaanse Inkomstediens (SARS), en die kategoriedeterminante, soos onderwysvlak en indiensnemingstatus, op belastingvoldoeningskoste het. Die CHAID-ontleding het dus ʼn bykomende vlak van insig voorsien wat nie moontlik is met regressieontleding nie en sodoende die bruikbaarheid van die resultate verbeter. Indiensnemingstatus en inkomstebelastingkategorie het die sterkste assosiasie met belastingvoldoeningskoste (belastingpligtiges in eie diens en belastingpligtiges in die hoogste inkomstekategorie het gemiddeld die hoogste totale belastingvoldoeningskoste). Verskeie ander determinante, byvoorbeeld, soort bystand verkry, geslag, onderwysvlak, kompleksiteit van belastingwetgewing, kompleksiteit van SARS-gidse en SARS-diensgehaltegradering is ook in die ontledings geïdentifiseer. Die studie het afgesluit met voorstelle om individuele belastingspligtiges se belastingvoldoeningskoste te verlaag. / Ditshenyegelo tša ditefelo tša go obamela melawana ya metšhelo ka balefamotšhelo ka Afrika Borwa ga di tsebje, gomme go bohlokwa gore re tsebe, gareng ga tše dingwe, ge eba ditshenyegelo tše di tliša kotsi ya go baka gore balefamotšhelo ba se ke ba obamela melawana ya metšhelo, e lego seo se ka bago le seabe sa go se loke go kgoboketšo ya letseno la motšhelo. Ke ka lebaka leo, maikemišetšo a dinyakišišo tše e bile go sekaseka ditshenyegelo tša go obamela melawana ya motšhelo ka balefamotšhelo ka Afrika Borwa. Tshekatsheko ya ditshenyegelo tša go obamela melawana ya motšhelo go ra gore re swanetše go hlakanya ditshenyegelo tše mabapi le go romela dingwalwa tša motšhelo le ditiragalo tša ka morago ga go romela dingwalwa tšeo tša motšhelo, go realo e le go tseba dilo tšeo di bakago ditshenyegelo tše le go šišinya mekgwa ya go di fokotša. Ka go šomiša tshedimošo ye e kgobokeditšwego go dinyakišišo tšeo di dirilwego ka inthanete gareng ga balefamotšhelo ba 10 260, go akantšwe gore (ka go diriša mekgwa ye e fapafapanego) ditshenyegelo tša go obamela melawana ya motšhelo wa letseno ka balefamotšhelo ka ngwaga wa tshekatsheko ya metšhelo wa 2018 di bile magareng ga 3.61% le 5.31% ya tšhelete ya motšhelo wa letseno. Dipoelo tše di bapetšwa gabotse le dikelo tšeo di begilwego ka dinyakišišong tše ntši tšeo di dirilwego ka dinageng tše dingwe gomme di laeditše go fokotšega go tšwa go dipoelo tšeo di hweditšwego go balefamotšhelo ba 752 ka ngwageng wa tshekatsheko ya metšhelo wa 2017. Kutollo ya tirišano ya maitirišo ya Chi-square (CHAID), e lego mokgwa wa go nyakišiša sephetho ka maphakga, e šomišitšwe ka nepo ya go tseba dilo tšeo di bakago ditshenyegelo tša go obamela melawana ya motšhelo le go hlaola dihlopha tše itšego tša balefamotšhelo bao ba amanago le mehuta ye e swanago e nnoši ya dihlaodi tšeo di bilego bohlokwa go ya ka dipalopalo mabapi le ditshenyegelo tša go obamela melawana ya motšhelo. Karoganyo ye e kgontšhitše kwešišo ye kaone ya khuetšo yeo dikokwane tše itšego tša dilo tšeo di tšwelago pele go baka se, tša go swana le kelo ya boleng bja tirelo ye e abjago ke Tirelo ya Motšhelo ya Afrika Borwa (SARS), le dilo tšeo di bakago go se obamele melawana ya motšhelo go ya ka makala, go swana le maemo a thuto le maemo a mošomo, go ditshenyegelo tša go obamela melawana ya motšhelo. Tshekatsheko ya CHAID ka fao e file maemo a tlaleletšo a tsebo yeo e sa kgonagalego ka tshekatsheko ya poelomorago, go maatlafatša go šomišega ga dipoelo. Maemo a mošomo le legoro la motšhelo wa letseno di bile le kamano ye kgolo le ditshenyegelo tša go obamela melawana ya motšhelo (ka kakaretšo, balefamotšhelo bao ba itšhomelago le balefamotšhelo bao ba lego ka go legoro la godimo la motšhelo wa letseno ba na le palomoka ya godimodimo ya motšhelo wo ba lefišwago ona). Dilo tše dingwe tša mehutahuta tšeo di bakago go se obamele melawana ya motšhelo, go fa mohlala, mohuta wa thušo ye e hweditšwego, bong, maemo a thuto, go se kwešišege ga melawana ya motšhelo, go se kwešišege ga ditlhahli tša SARS, le kelo ya boleng bja tirelo ya SARS le tšona di utollotšwe ka tshekatshekong. Dinyakišišo di feditše ka go fa ditšhišinyo mabapi le ka fao go ka fokotšwago ditshenyegelo tša go obamela melawana ya motšhelo ka balefamotšhelo. / D. Phil. (Accounting Sciences)