La notion de biomarqueurs est définie comme « une caractéristique mesurée objectivement et évaluée comme indicateur de processus biologiques normaux ou pathologiques, ou de réponses pharmacologiques à une intervention thérapeutique ». L'intérêt scientifique pour les biomarqueurs est de plus en plus important. Ils permettent, entre autres,une meilleure compréhension des processus pathologiques et de diagnostiquer, voire pronostiquer ces pathologies. Les études « omiques » telles que la lipidomique jouent un rôle essentiel dans la découverte de nouveaux biomarqueurs. La lipidomique consiste à explorer le lipidome d'un échantillon biologique et à déceler l'impact de la pathologie sur ce dernier. Les lipides constituent une vaste et importante famille de métabolites retrouvés dans toutes les cellules vivantes, dont leur nombre est estimé à plus de 100 000 espèces chez les mammifères. Ils sont impliqués, notamment, dans le stockage d'énergie et la transduction de signal. Mon travail de thèse a reposé sur la réalisation d'approches lipidomiques en LC-MS sur diverses applications en santé telles que le syndrome de déficit immunitaire combiné sévère associé à une alopécie et une dystrophie des ongles, le syndrome du nystagmus infantile et le rejet de greffe rénale. A cette fin, des analyses statistiques multivariées et univariées ont été employées pour déceler des potentiels lipides biomarqueurs. / Biomarker was defined as "a characteristic that is objectively measured and evaluated as an indicator of normal biological processes, pathogenic processes, or pharmacological responses to therapeutic intervention". The scientific interest in biomarkers is more and more important. They allow, in particular, to better understand pathogenic processes and to diagnose, even to predict pathologies. "Omics" studies, such as lipidomics, play an essential role in the new biomarkers discovery. Lipidomics consist in exploring biological samples lipidome and in detecting pathogenic impact on this latter. Lipids are a large and important metabolite family found in all living cells. Their quantity is estimated to more than 100,000 species in mammals. They are involved, in particular, in the energy storage and the signal transduction. My PhD thesis involved carrying out lipidomics approaches with LC-MS through various health applications such as severe combined immunodeficiency associated with alopecia syndrome, infantile nystagmus syndrome and renal graft rejection. For this purpose, multivariate and univariate statistical analyses were carried out in order to detect potential lipid biomarkers.

O objetivo principal do presente trabalho é oferecer métodos alternativos de previsão da arrecadação tributária federal, baseados em metodologias de séries temporais, inclusive com a utilização de variáveis explicativas, que reflitam a influência do cenário macroeconômico na arrecadação tributária, com o intuito de melhorar a acurácia da previsão da arrecadação. Para tanto, foram aplicadas as metodologias de modelos dinâmicos univariados, multivariados, quais sejam, Função de Transferência, Auto-regressão Vetorial (VAR), VAR com correção de erro (VEC), Equações Simultâneas, e de modelos Estruturais. O trabalho tem abrangência regional e limita-se à análise de três séries mensais da arrecadação, relativas ao Imposto de Importação, Imposto Sobre a Renda das Pessoas Jurídicas e Contribuição para o Financiamento da Seguridade Social - Cofins, no âmbito da jurisdição do estado de São Paulo, no período de 2000 a 2007. Os resultados das previsões dos modelos acima citados são comparados entre si, com a modelagem ARIMA e com o método dos indicadores, atualmente utilizado pela Secretaria da Receita Federal do Brasil (RFB) para previsão anual da arrecadação tributária, por meio da raiz do erro médio quadrático de previsão (RMSE). A redução média do RMSE foi de 42% em relação ao erro cometido pelo método dos indicadores e de 35% em relação à modelagem ARIMA, além da drástica redução do erro anual de previsão. A utilização de metodologias de séries temporais para a previsão da arrecadação de receitas federais mostrou ser uma alternativa viável ao método dos indicadores, contribuindo para previsões mais precisas, tornando-se ferramenta segura de apoio para a tomada de decisões dos gestores. / The main objective of this work is to offer alternative methods for federal tax revenue forecasting, based on methodologies of time series, inclusively with the use of explanatory variables, which reflect the influence of the macroeconomic scenario in the tax collection, for the purpose of improving the accuracy of revenues forecasting. Therefore, there were applied the methodologies of univariate dynamic models, multivariate, namely, Transfer Function, Vector Autoregression (VAR), VAR with error correction (VEC), Simultaneous Equations, and Structural Models. The work has a regional scope and it is limited to the analysis of three series of monthly tax collection of the Import Duty, the Income Tax Law over Legal Entities Revenue and the Contribution for the Social Security Financing Cofins, under the jurisdiction of the state of São Paulo in the period from 2000 to 2007. The results of the forecasts from the models above were compared with each other, with the ARIMA moulding and with the indicators method, currently used by the Secretaria da Receita Federal do Brasil (RFB) to annual foresee of the tax collection, through the root mean square error of approximation (RMSE). The average reduction of RMSE was 42% compared to the error committed by the method of indicators and 35% of the ARIMA model, besides the drastic reduction in the annual forecast error. The use of time-series methodologies to forecast the collection of federal revenues has proved to be a viable alternative to the method of indicators, contributing for more accurate predictions, becoming a safe support tool for the managers decision making process.

Nesse trabalho foi abordada a situação em que observações de produtividade da cana-de-açúcar (TCH) foram tomadas na mesma unidade experimental em diferentes condições de avalições (anos). Foram avaliados os perfis médios de resposta de 48 genótipos de cana-de-açúcar em dois experimentos: Experimento 1 e Experimento 2, durante três e cinco anos respectivamente, ambos com o delineamento de blocos ao acaso. Esse tipo de planejamento produz uma forma de relação entre as observações tomadas na mesma unidade experimental, portanto requer outras suposições, além das usuais, para que análise seja correta e os testes produzam resultados válidos. Para que as inferências sobre as médias de produtividade sejam válidas e seguras é necessário que o modelo da matriz de covariância dos dados seja apropriado. Diante disso, foram avalidos três alterantivas de análise para dados longitudinais (medidas repetidas no tempo ), sendo utilizados portanto, o modelo univariado, conforme o planejamento do tipo \"split-plot on time\", que impõe forte restrição quanto a matriz de variâncias-covariâncias; o modelo multivariado, que utiliza uma matriz de variâncias-covariâncias não-estruturada e o modelo mistos, que possibilita a seleção de uma matriz que melhor representa os dados. Contudo, verificou-se que não houve diferença entre os resultados dos testes para as diferentes metodologias. Porém, é interessante a continuidade do estudo em relação ao modelo misto, pois devido a sua flexibilidade e precisão é possível obter estimativas mais seguras dos componentes de variância e predizer os valores genotípicos, que por fim poderá proporcionar a predição de produção de uma futura colheita para um determinado genótipo. / This work has been dealt with situation in which observations of productivity of sugar of cane (TCH) were taken in the same unit experimental in different condition of assessments (years). The response profiles average of 48 genotypes of sugar of cane were evaluated in two experiments: Experiment 1 and Experiment 2, for three and five years respectively, both with the randomized complete block design. This type of planning produces a form of relationship between the observations made in the same unit experimental therefore requires other assumptions, in addition to the usual, so that analysis is correct and the test results valid. To that inferences on the means of productivity are valid and safe it is necessary that the model of covariance matrix of the data is appropriate. Therefore, were evaluated three alternatives for analysis of longitudinal data (repeated measures over time), the univariate model as the planning of the split-plot on time which imposes strong restrictions on variances - covariances matrix, the multivariate model, which uses a non-structured variances - covariances matrix and mixed model, which they are enable the selection of a matrix that best represents the data. However, it was found that there was no difference between the results of tests for the different methodologies. But it is interesting the continuity of the study in relation to mixed model, because due to its flexibility and accuracy will be possible to obtain more reliable estimates of the variance components and predict the genotypic values, which ultimately could provide a prediction of production of a future harvest for a given genotype.

[pt] O ONS (operador nacional do sistema elétrico brasileiro) vem utilizando o software ANNSTLF produzido pelo EPRI/EUA (Eletrical Power Research Institute) para realizar a previsão do consumo de carga horária. Entretanto, as estimativas fornecidas pelo programa estão fundamentadas na metodologia de uma rede neural que, de certo modo, impede ao usuário de extrair uma maior interpretação dos resultados que são fornecidos pela rede. Assim sendo, este trabalho pesquisou os métodos univariados convencionais: Holt-Winters e Box e Jenkins, considerando suas formulações aperfeiçoadas e adaptadas às características próprias do tipo de série em questão. Isto é, assumindo a existência de dois ciclos sazonais: um diário e outro semanal. A vantagem destas técnicas univariadas, em comparação ao ANNSTLF, é principalmente a interpretabilidade das informações obtidas. Dessa forma, esta pesquisa permite também avaliar melhor o desempenho do ANNSTLF. / [en] The ONS (National Operator of the Brazilian electrical system) has been using the software ANNSTLF produced by EPRI/USA (Eletrical Power Research Institute) to carry out the forecast of the hourly load consumption. However, the estimates supplied by the program are based on the methodology of a neural net that, in a way, does not allow the user to extract a better interpretation of the results produced by the net. Therefore, investigates the conventional univaried methods: Holt-Winters and Box & Jenkins, considering its formulations perfected and adapted to the characteristics of the series understudy. That is, its assumed the existence of two seasonal cicles: daily and weekly. The advantage of these univariate techniques, in comparison to the ANNSTLF, is mainly the ability to interpret the model estimates. Also, this research also allows a better evaluation the performance of the ANNSTLF.

Submitted by Christopher Davies Junior (christopher.junior@volkswagen.com.br) on 2011-09-05T20:59:11Z No. of bitstreams: 1 Utilizando modelos econométricos univariados.pdf: 443280 bytes, checksum: c9deb332599852aebed4578c3453998c (MD5) / Approved for entry into archive by Gisele Isaura Hannickel (gisele.hannickel@fgv.br) on 2011-09-06T12:24:10Z (GMT) No. of bitstreams: 1 Utilizando modelos econométricos univariados.pdf: 443280 bytes, checksum: c9deb332599852aebed4578c3453998c (MD5) / Approved for entry into archive by Gisele Isaura Hannickel (gisele.hannickel@fgv.br) on 2011-09-06T12:24:21Z (GMT) No. of bitstreams: 1 Utilizando modelos econométricos univariados.pdf: 443280 bytes, checksum: c9deb332599852aebed4578c3453998c (MD5) / Made available in DSpace on 2011-09-06T12:29:13Z (GMT). No. of bitstreams: 1 Utilizando modelos econométricos univariados.pdf: 443280 bytes, checksum: c9deb332599852aebed4578c3453998c (MD5) Previous issue date: 2011-08-18 / A previsão de demanda é uma atividade relevante pois influencia na tomada de decisão das organizações públicas e privadas. Este trabalho procura identificar modelos econométricos que apresentem bom poder preditivo para a demanda automotiva brasileira num horizonte de longo prazo, cinco anos, através do uso das séries de vendas mensais de automóveis, veículos comerciais leves e total, o período amostral é de 1970 a 2010. Foram estimados e avaliados os seguintes modelos: Auto-regressivo (Box-Jenkins, 1976), Estrutural (Harvey, 1989) e Mudança de Regime (Hamilton, 1994), incluindo efeitos calendário e dummies além dos testes de raízes unitárias sazonais e não-sazonais para as séries. A definição da acurácia dos modelos baseou-se no Erro Quadrático Médio (EQM) dos resultados apresentados na simulação da previsão de demanda dos últimos quinze anos (1995 a 2010). / Demand forecasting is an important activity because it influences the decision making of public and private organizations. This study identifies econometrics models that have good predictive power for the Brazilian automotive demand in a long-term horizon, five years, through the use of series of monthly sales of automobiles, light commercial vehicles and total, the sample period is from 1970 until 2010. Were estimated and evaluated the following models: Auto-regressive (Box-Jenkins, 1976), Structural (Harvey, 1989) and Regime Change (Hamilton, 1994), including calendar effects and dummies, also seasonal and non-seasonal unit root tests for the series. The definition of the accuracy of the models was based on Mean Square Error (MSE) of the results presented in the simulation of the forecast demand over the last fifteen years (1995 till 2010).

O objetivo principal do presente trabalho é oferecer métodos alternativos de previsão da arrecadação tributária federal, baseados em metodologias de séries temporais, inclusive com a utilização de variáveis explicativas, que reflitam a influência do cenário macroeconômico na arrecadação tributária, com o intuito de melhorar a acurácia da previsão da arrecadação. Para tanto, foram aplicadas as metodologias de modelos dinâmicos univariados, multivariados, quais sejam, Função de Transferência, Auto-regressão Vetorial (VAR), VAR com correção de erro (VEC), Equações Simultâneas, e de modelos Estruturais. O trabalho tem abrangência regional e limita-se à análise de três séries mensais da arrecadação, relativas ao Imposto de Importação, Imposto Sobre a Renda das Pessoas Jurídicas e Contribuição para o Financiamento da Seguridade Social - Cofins, no âmbito da jurisdição do estado de São Paulo, no período de 2000 a 2007. Os resultados das previsões dos modelos acima citados são comparados entre si, com a modelagem ARIMA e com o método dos indicadores, atualmente utilizado pela Secretaria da Receita Federal do Brasil (RFB) para previsão anual da arrecadação tributária, por meio da raiz do erro médio quadrático de previsão (RMSE). A redução média do RMSE foi de 42% em relação ao erro cometido pelo método dos indicadores e de 35% em relação à modelagem ARIMA, além da drástica redução do erro anual de previsão. A utilização de metodologias de séries temporais para a previsão da arrecadação de receitas federais mostrou ser uma alternativa viável ao método dos indicadores, contribuindo para previsões mais precisas, tornando-se ferramenta segura de apoio para a tomada de decisões dos gestores. / The main objective of this work is to offer alternative methods for federal tax revenue forecasting, based on methodologies of time series, inclusively with the use of explanatory variables, which reflect the influence of the macroeconomic scenario in the tax collection, for the purpose of improving the accuracy of revenues forecasting. Therefore, there were applied the methodologies of univariate dynamic models, multivariate, namely, Transfer Function, Vector Autoregression (VAR), VAR with error correction (VEC), Simultaneous Equations, and Structural Models. The work has a regional scope and it is limited to the analysis of three series of monthly tax collection of the Import Duty, the Income Tax Law over Legal Entities Revenue and the Contribution for the Social Security Financing Cofins, under the jurisdiction of the state of São Paulo in the period from 2000 to 2007. The results of the forecasts from the models above were compared with each other, with the ARIMA moulding and with the indicators method, currently used by the Secretaria da Receita Federal do Brasil (RFB) to annual foresee of the tax collection, through the root mean square error of approximation (RMSE). The average reduction of RMSE was 42% compared to the error committed by the method of indicators and 35% of the ARIMA model, besides the drastic reduction in the annual forecast error. The use of time-series methodologies to forecast the collection of federal revenues has proved to be a viable alternative to the method of indicators, contributing for more accurate predictions, becoming a safe support tool for the managers decision making process.

Nesse trabalho foi abordada a situação em que observações de produtividade da cana-de-açúcar (TCH) foram tomadas na mesma unidade experimental em diferentes condições de avalições (anos). Foram avaliados os perfis médios de resposta de 48 genótipos de cana-de-açúcar em dois experimentos: Experimento 1 e Experimento 2, durante três e cinco anos respectivamente, ambos com o delineamento de blocos ao acaso. Esse tipo de planejamento produz uma forma de relação entre as observações tomadas na mesma unidade experimental, portanto requer outras suposições, além das usuais, para que análise seja correta e os testes produzam resultados válidos. Para que as inferências sobre as médias de produtividade sejam válidas e seguras é necessário que o modelo da matriz de covariância dos dados seja apropriado. Diante disso, foram avalidos três alterantivas de análise para dados longitudinais (medidas repetidas no tempo ), sendo utilizados portanto, o modelo univariado, conforme o planejamento do tipo \"split-plot on time\", que impõe forte restrição quanto a matriz de variâncias-covariâncias; o modelo multivariado, que utiliza uma matriz de variâncias-covariâncias não-estruturada e o modelo mistos, que possibilita a seleção de uma matriz que melhor representa os dados. Contudo, verificou-se que não houve diferença entre os resultados dos testes para as diferentes metodologias. Porém, é interessante a continuidade do estudo em relação ao modelo misto, pois devido a sua flexibilidade e precisão é possível obter estimativas mais seguras dos componentes de variância e predizer os valores genotípicos, que por fim poderá proporcionar a predição de produção de uma futura colheita para um determinado genótipo. / This work has been dealt with situation in which observations of productivity of sugar of cane (TCH) were taken in the same unit experimental in different condition of assessments (years). The response profiles average of 48 genotypes of sugar of cane were evaluated in two experiments: Experiment 1 and Experiment 2, for three and five years respectively, both with the randomized complete block design. This type of planning produces a form of relationship between the observations made in the same unit experimental therefore requires other assumptions, in addition to the usual, so that analysis is correct and the test results valid. To that inferences on the means of productivity are valid and safe it is necessary that the model of covariance matrix of the data is appropriate. Therefore, were evaluated three alternatives for analysis of longitudinal data (repeated measures over time), the univariate model as the planning of the split-plot on time which imposes strong restrictions on variances - covariances matrix, the multivariate model, which uses a non-structured variances - covariances matrix and mixed model, which they are enable the selection of a matrix that best represents the data. However, it was found that there was no difference between the results of tests for the different methodologies. But it is interesting the continuity of the study in relation to mixed model, because due to its flexibility and accuracy will be possible to obtain more reliable estimates of the variance components and predict the genotypic values, which ultimately could provide a prediction of production of a future harvest for a given genotype.

Submitted by Renata Lopes (renatasil82@gmail.com) on 2017-03-02T12:24:39Z No. of bitstreams: 1 annaclaudiamancinidasilvacarneiro.pdf: 1333903 bytes, checksum: a7b3819bb5b0e1adb8efd07bca0f9aa2 (MD5) / Approved for entry into archive by Adriana Oliveira (adriana.oliveira@ufjf.edu.br) on 2017-03-06T19:35:55Z (GMT) No. of bitstreams: 1 annaclaudiamancinidasilvacarneiro.pdf: 1333903 bytes, checksum: a7b3819bb5b0e1adb8efd07bca0f9aa2 (MD5) / Made available in DSpace on 2017-03-06T19:35:55Z (GMT). No. of bitstreams: 1 annaclaudiamancinidasilvacarneiro.pdf: 1333903 bytes, checksum: a7b3819bb5b0e1adb8efd07bca0f9aa2 (MD5) Previous issue date: 2014-09-26 / CAPES - Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / A previsão de cargas elétricas é fundamental para o planejamento das empresas de energia. O foco deste estudo são as previsões a curto prazo; assim, aplicamos métodos univariados de previsão de séries temporais a uma série real de cargas elétricas de 104 semanas no Rio de Janeiro, nos anos de 1996 e 1997, e experimentamos várias combinações dos métodos de melhor desempenho. As combinações foram feitas pelo método outperformance, uma combinação linear simples, com pesos fixos. Os resultados das combinações foram comparados ao de simulações de redes neurais artificiais que solucionam o mesmo problema, e ao resultado de um método de amortecimento de dupla sazonalidade aditiva. No geral, este método de amortecimento obteve os melhores resultados, e talvez seja o mais adequado e confiável para aplicações práticas, embora necessite de melhorias para garantir a extração completa da informação contida nos dados. / Forecasting the demand for electric power is crucial for the production planning in energy utilities. The focus of this study are the short-term forecasts. We apply univariate time series methods to the forecasting of a series containing observations of the energy consumption of 104 weeks in Rio de Janeiro, in 1996 and 1997, and experiment with several combinations of the methods which have the best performance. These combinations are done by the outperformance method, a simple linear combination with fixed weights. The results were compared to those obtained by neural networks on the same problem, and with the results of a exponential smoothing method for dual additive seasonality. Overall, the exponential smoothing method achieved the best results, and was shown to be perhaps the most reliable and suitable for practical applications, even though it needs improvements to ensure complete extraction of the information contained in the data.

La limite des connaissances techniques ou fondamentale, est une réalité dont l’industrie fait face. Le besoin de mettre à jour cette connaissance acquise est essentiel pour une compétitivité économique, mais aussi pour une meilleure maniabilité des systèmes et machines. Aujourd’hui grâce à ces systèmes et machine, l’expansion de données en quantité, en fréquence de génération est un véritable phénomène. À présent par exemple, les avions Airbus génèrent des centaines de mégas de données par jour, et intègrent des centaines voire des milliers de capteurs dans les nouvelles générations d’avions. Ces données générées par ces capteurs, sont exploitées au sol ou pendant le vol, pour surveiller l’état et la santé de l’avion, et pour détecter des pannes, des incidents ou des changements. En théorie, ces pannes, ces incidents ou ces changements sont connus sous le terme d’anomalie. Une anomalie connue comme un comportement qui ne correspond pas au comportement normal des données. Certains la définissent comme une déviation d’un modèle normal, d’autres la définissent comme un changement. Quelques soit la définition, le besoin de détecter cette anomalie est important pour le bon fonctionnement de l'avion. Actuellement, la détection des anomalies à bord des avions est assuré par plusieurs équipements de surveillance aéronautiques, l’un de ces équipements est le « Aircraft condition monitoring System –ACMS », enregistre les données générées par les capteurs en continu, il surveille aussi l’avion en temps réel grâce à des triggers et des seuils programmés par des Airlines ou autres mais à partir d’une connaissance a priori du système. Cependant, plusieurs contraintes limitent le bon fonctionnement de cet équipement, on peut citer par exemple, la limitation des connaissances humaines un problème classique que nous rencontrons dans plusieurs domaines. Cela veut dire qu’un trigger ne détecte que les anomalies et les incidents dont il est désigné, et si une nouvelle condition surgit suite à une maintenance, changement de pièce, etc. Le trigger est incapable s’adapter à cette nouvelle condition, et il va labéliser toute cette nouvelle condition comme étant une anomalie. D’autres problèmes et contraintes seront cités progressivement dans les chapitres qui suivent. L’objectif principal de notre travail est de détecter les anomalies et les changements dans les données de capteurs, afin d’améliorer le system de surveillance de santé d’avion connu sous le nom Aircraft Health Monitoring(AHM). Ce travail est basé principalement sur une analyse à deux étapes, Une analyse unie varie dans un contexte non supervisé, qui nous permettra de se focaliser sur le comportement de chaque capteur indépendamment, et de détecter les différentes anomalies et changements pour chaque capteur. Puis une analyse multi-variée qui nous permettra de filtrer certaines anomalies détectées (fausses alarmes) dans la première analyse et de détecter des groupes de comportement suspects. La méthode est testée sur des données réelles et synthétiques, où les résultats, l’identification et la validation des anomalies sont discutées dans cette thèse. / The limitation of the knowledge, technical, fundamental is a daily challenge for industries. The need to updates these knowledge are important for a competitive industry and also for an efficient reliability and maintainability of the systems. Actually, thanks to these machines and systems, the expansion of the data on quantity and frequency of generation is a real phenomenon. Within Airbus for example, and thanks to thousands of sensors, the aircrafts generate hundreds of megabytes of data per flight. These data are today exploited on the ground to improve safety and health monitoring system as a failure, incident and change detection. In theory, these changes, incident and failure are known as anomalies. An anomaly is known as deviation form a normal behavior of the data. Others define it as a behavior that do not conform the normal behavior. Whatever the definition, the anomaly detection process is very important for good functioning of the aircraft. Currently, the anomaly detection process is provided by several health monitoring equipments, one of these equipment is the Aircraft Health Monitoring System (ACMS), it records continuously the date of each sensor, and also monitor these sensors to detect anomalies and incident using triggers and predefined condition (exeedance approach). These predefined conditions are programmed by airlines and system designed according to a prior knowledge (physical, mechanical, etc.). However, several constraints limit the ACMS anomaly detection potential. We can mention, for example, the limitation the expert knowledge which is a classic problem in many domains, since the triggers are designed only to the targeted anomalies. Otherwise, the triggers do not cover all the system conditions. In other words, if a new behavior appears (new condition) in the sensor, after a maintenance action, parts changing, etc. the predefined conditions won't detect any thing and may be in many cases generated false alarms. Another constraint is that the triggers (predefined conditions) are static, they are unable to adapt their proprieties to each new condition. Another limitation is discussed gradually in the future chapters. The principle of objective of this thesis is to detect anomalies and changes in the ACMS data. In order to improve the health monitoring function of the ACMS. The work is based principally on two stages, the univariate anomaly detection stage, where we use the unsupervised learning to process the univariate sensors, since we don’t have any a prior knowledge of the system, and no documentation or labeled classes are available. The univariate analysis focuses on each sensor independently. The second stage of the analysis is the multivariate anomaly detection, which is based on density clustering, where the objective is to filter the anomalies detected in the first stage (false alarms) and to detect suspected behaviours (group of anomalies). The anomalies detected in both univariate and multivariate can be potential triggers or can be used to update the existing triggers. Otherwise, we propose also a generic concept of anomaly detection based on univariate and multivariate anomaly detection. And finally a new concept of validation anomalies within airbus.

In this thesis, we explored the generative adversarial network called uTSGAN to generate patterns from multivariate CAN bus time series dataset. Since the given data is unlabelled, unprocessed and highly imbalanced containing large amount of missing values, we have to define and discard a few timestamps and limit the focus of the study to the reduced subset involving patterns of the 10-second window size, which are categorised and clustered into majority and minority classes. To generate such an imbalanced set, we have used image based time series GAN called uTSGAN which transforms a time sequence into a spectrogram image and back to a time sequence within the same GAN framework. For the comparison, we also prepared a resampled (balanced) dataset from the imbalanced set to use in alternative experiments. This comparison evaluates the results of the conventional resampling approach against the baseline as well as our novel implementations. We propose two new methods using "cluster density based" and "sample loss based" techniques. Throughout the experimentation, the "cluster density based" GANs consistently achieved better results on some common and uncommon evaluations for multivariate and highly imbalanced sample sets. Among some regular evaluation methods, classification metrics such as balanced accuracy and precision provide a better understanding of experimentation results. The TRTS balanced accuracy and precision from "cluster density based" GAN achieves over 82% and 90% with an improvement of 20-30% and 14-18% respectively from that of baseline; the TSTR balanced accuracy of "cluster density based" increased by 10.6% from that of baseline and it show slightly better precision with respect to that of the baseline when compared on generated results from univariate experiments. Secondly, the alternative "resampling based" implementations show similar values to that of the baseline in TRTS and TSTR classifications. Simultaneously, More distinguished results are seen using a quantitative metric called Earth Mover’s Distance(EMD). We have used this distance measure to calculate the overall mean EMD and clusterwise mean EMD between real samples and fake (i.e. generated) samples. During their evaluations, "cluster density based" experiments showed significantly better results for not only majority but also minority clusters as compared to the results of baseline and "resampling based" experiments. At the end, we have opened a discussion on how one can utilize our findings in MAR problem aswell as improve the results by taking some precautionary measures.