An investigation was made into the characteristics of program elements and removals of Weapon Replaceable Assemblies aggregated at the system level for the purpose of developing a method to predict removals during initial provisioning periods. From examination of nine avionic systems over a 28 month period a binomial model was developed using a removal rate based on aircraft-months as a program element. The model is to be used before Fleet data are generated by obtaining aircraft-month estimates from the contractor and removal rate estimates from similar operational equipments. A probability distribution reflecting the degree of certainty is selected as a prior estimate. Then, as Fleet experience is accumulated the distribution is updated using Bayesian techniques and maturity growth curves. This distribution is used to give an estimate of current removal rate and to extrapolate to future removal rates.

Current works of bike demand prediction mainly focus on cluster level and perform poorly on predicting demands of a single station. In the first task, we introduce a contextual based bike demand prediction model, which predicts bike demands for per station by combining spatio-temporal network and environment contexts synergistically. Furthermore, since people's movement information is an important factor, which influences the bike demands of each station. To have a better understanding of people's movements, we need to analyze the relationship between different places. In the second task, we propose an origin-destination model to learn place representations by using large scale movement data. Then based on the people's movement information, we incorporate the place embedding into our bike demand prediction model, which is built by using multi-source large scale datasets: New York Citi bike data, New York taxi trip records, and New York POI data. Finally, as deep learning methods have been successfully applied to many fields such as image recognition and natural language processing, it inspires us to incorporate the complex deep learning method into the bike demand prediction problem. So in this task, we propose a deep spatial-temporal (DST) model, which contains three major components: spatial dependencies, temporal dependencies, and external influence. Experiments on the NYC Citi Bike system show the effectiveness and efficiency of our model when compared with the state-of-the-art methods.

O presente trabalho propõe uma evolução metodológica na operação do Sistema Adutor Metropolitano de São Paulo, em tempo real. Foi implantado um modelo matemático, em tempo real, de previsão de consumo de água horário para uma melhoria na performance operacional. Descrevem-se vários procedimentos de sistema de controle operacional, desde manual até totalmente automático, em sistemas de abastecimento. O sistema de abastecimento de São Paulo é classificado neste contexto. Foi analisada a possibilidade de desenvolvimento da situação atual rumo a um controle mais eficiente, através do uso de um modelo de previsão de demanda de água. O “estado da arte" em modelos de previsão de consumo de água é apresentado através de uma revisão bibliográfica especifica. Foi desenvolvida uma interface entre um modelo de rede hidráulica e um modelo de previsão de demanda de água existente, ambos utilizando dados operacionais, obtidos em tempo real de um sistema de telemetria. A interface foi testada em um estudo de caso do Sistema Adutor de São Paulo. Com a utilização de um modelo de previsão, concluiu-se que é possível estabelecer regras operacionais mais eficientes. Essa eficiência é demonstrada pela redução do número de mudanças de posição de válvula e estado de bombas, bem como é observada a redução do custo de energia elétrica (reduzindo o bombeamento em horário de maior custo). Os benefícios obtidos do uso conjunto do modelo simulador hidráulico e do modelo de previsão de demanda não podem ser considerados como o ótimo global. Seria necessário dispor de um modelo de otimização (programação automática). De qualquer forma, foi concluído que o investimento na implementação desses dois modelos é extremamente atrativa. / This work proposes a methodological evolution of a real time water distribution system operation applied to the Water Mains System of Metropolitan Region of Sao Paulo. It was settled a mathematical model in real time, to forecast hourly water consumptions, intending to increase operational performance. Several operational control procedures of water systems were described, since manual ones until total automatic ones. Sao Paulo system is classified into this concept. The possibility of development from the present status toward a more efficient control was analyzed, through the use of a water demand prediction model. State-of-art of water demand models is presented, through a specific literature review. An interface between a hydraulic network model and an existing water demand prediction model were developed both of them using operational data, obtained in real time by a telemetric system. The interface was tested in a case study of Sao Paulo Water Mains System. One concludes that through the use of the prediction model, it was possible to make more efficient operational schedules. This efficiency is demonstrated by the reduction in number of valve positions changes and in pump status changes, as well as a decrease in energy costs could be observed ( reducing pump operations in hours of more expensive costs). Benefits obtained by the conjunctive use of the hydraulic simulation model and the water demand prediction model can not be admitted as the global optimum. It would be necessary to make available an optimization model (automatic scheduler). However it was concluded that investment in these two models implementations is extremely attractive.

O presente trabalho propõe uma evolução metodológica na operação do Sistema Adutor Metropolitano de São Paulo, em tempo real. Foi implantado um modelo matemático, em tempo real, de previsão de consumo de água horário para uma melhoria na performance operacional. Descrevem-se vários procedimentos de sistema de controle operacional, desde manual até totalmente automático, em sistemas de abastecimento. O sistema de abastecimento de São Paulo é classificado neste contexto. Foi analisada a possibilidade de desenvolvimento da situação atual rumo a um controle mais eficiente, através do uso de um modelo de previsão de demanda de água. O “estado da arte” em modelos de previsão de consumo de água é apresentado através de uma revisão bibliográfica especifica. Foi desenvolvida uma interface entre um modelo de rede hidráulica e um modelo de previsão de demanda de água existente, ambos utilizando dados operacionais, obtidos em tempo real de um sistema de telemetria. A interface foi testada em um estudo de caso do Sistema Adutor de São Paulo. Com a utilização de um modelo de previsão, concluiu-se que é possível estabelecer regras operacionais mais eficientes. Essa eficiência é demonstrada pela redução do número de mudanças de posição de válvula e estado de bombas, bem como é observada a redução do custo de energia elétrica (reduzindo o bombeamento em horário de maior custo). Os benefícios obtidos do uso conjunto do modelo simulador hidráulico e do modelo de previsão de demanda não podem ser considerados como o ótimo global. Seria necessário dispor de um modelo de otimização (programação automática). De qualquer forma, foi concluído que o investimento na implementação desses dois modelos é extremamente atrativa. / This work proposes a methodological evolution of a real time water distribution system operation applied to the Water Mains System of Metropolitan Region of Sao Paulo. It was settled a mathematical model in real time, to forecast hourly water consumptions, intending to increase operational performance. Several operational control procedures of water systems were described, since manual ones until total automatic ones. Sao Paulo system is classified into this concept. The possibility of development from the present status toward a more efficient control was analyzed, through the use of a water demand prediction model. State-of-art of water demand models is presented, through a specific literature review. An interface between a hydraulic network model and an existing water demand prediction model were developed both of them using operational data, obtained in real time by a telemetric system. The interface was tested in a case study of Sao Paulo Water Mains System. One concludes that through the use of the prediction model, it was possible to make more efficient operational schedules. This efficiency is demonstrated by the reduction in number of valve positions changes and in pump status changes, as well as a decrease in energy costs could be observed ( reducing pump operations in hours of more expensive costs). Benefits obtained by the conjunctive use of the hydraulic simulation model and the water demand prediction model can not be admitted as the global optimum. It would be necessary to make available an optimization model (automatic scheduler). However it was concluded that investment in these two models implementations is extremely attractive.

To run a district heating system as efficiently as possible correct unit-commitmentdecisions has to be made and in order to make those decisions a good forecast ofheat demand for the coming planning period is necessary. With a high quality forecastthe need for backup power and the risk for a too high production are lowered. Thisthesis takes a neural network approach to load forecasting and aims to provide asimple, yet powerful, tool that can provide accurate load forecasts from existingproduction data without the need for extensive model building.The developed software is tested using real life data from two co-generation plantsand the conclusion is that when the quality of the raw data is good, the software canproduce very good forecasting results.

Allt eftersom klimatkrisen fortskrider ökar engagemanget kring hållbarhet inom företag. Växthusgaser är ett av de största problemen och matsvinn har därför fått mycket uppmärksamhet sedan det utnämndes till den tredje största bidragaren till de globala utsläppen. För att minska sitt bidrag rabatterar många matbutiker produkter med kort bästföredatum, vilket kommit att kräva en förståelse för hur priskänslig efterfrågan på denna typ av produkt är. Prisoptimering görs vanligtvis med så kallade Generalized Linear Models men då efterfrågan är ett komplext koncept har maskininl ärningsmetoder börjat utmana de traditionella modellerna. En sådan metod är Random Forest Regression, och syftet med uppsatsen är att utreda ifall modellen är bättre på att estimera efterfrågan baserat på rabattnivå än en klassisk linjär modell. Vidare utreds det ifall ett tydligt linjärt samband existerar mellan rabattnivå och efterfrågan, samt ifall detta beror av produkttyp. Resultaten visar på att Random Forest tar bättre hänsyn till det komplexa samband som visade sig finnas, och i detta specifika fall presterar bättre. Vidare visade resultaten att det sammantaget inte finns något linjärt samband, men att vissa produktkategorier uppvisar svag linjäritet. / As the climate crisis continues to evolve many companies focus their development on becoming more sustainable. With greenhouse gases being highlighted as the main problem, food waste has obtained a great deal of attention after being named the third largest contributor to global emissions. One way retailers have attempted to improve is through offering close-to-date produce at discount, hence decreasing levels of food being thrown away. To minimize waste the level of discount must be optimized, and as the products can be seen as flawed the known price-to-demand relation of the products may be insufficient. The optimization process historically involves generalized linear regression models, however demand is a complex concept influenced by many factors. This report investigates whether a Machine Learning model, Random Forest Regression, is better at estimating the demand of close-to-date products at different discount levels than a basic linear regression model. The discussion also includes an analysis on whether discounts always increase the will to buy and whether this depends on product type. The results show that Random Forest to a greater extent considers the many factors influencing demand and is superior as a predictor in this case. Furthermore it was concluded that there is generally not a clear linear relation however this does depend on product type as certain categories showed some linearity.

O presente trabalho propõe um modelo de operação sustentado por um sistema de suporte à decisão para operar a distribuição de água em tempo real atendendo a condições / restrições hidráulicas com o mínimo custo de energia elétrica. O atendimento às condições / restrições hidráulicas são avaliadas por um modelo simulador hidráulico previamente montado e calibrado. O conjunto de resultados avaliados pelo modelo de simulação hidráulica é analisado por um modelo de otimização proposto com solução de programação linear. As condições de operação em tempo real geram a necessidade de alimentação de informações operacionais automáticas a qualquer momento e com curto espaço de tempo – menor que horário. Para uma operação otimizada, previamente analisada por um modelo de simulação hidráulica cria uma condição critérios para uma previsão do consumo a ser atendido nas próximas horas. Um refinamento desses critérios são utilizados em um modelo de previsão de demanda de água que prevê e checa seus resultados de forma dinâmica. O modelo de operação proposto cria uma interface entre todos esses sistemas. Essa interface é testada e avaliada a partir de um estudo de caso aplicado no Sistema Adutor Metropolitano de São Paulo. A eficiência do modelo de operação proposto é apresentada tendo como resultado uma redução no custo de energia elétrica. / This assignment considers an operation model supported by decision support systems to operates the water supply systems in real time, considering the hydraulical conditions while achieving some performance goals, in this case, reducing electricity costs (minimization of pumping costs) the attempt of the hydraulic constraints are evaluated by an hydraulical simulator previously calibrated. The set of results are analyzed by an optimization model which uses a linear programming. The operation conditions in real time requires automatic feeding operational information shortly at any time (less than an hour) for an optimized operation, previously analyzed by a hydraulic simulation model with creates condition criteria of consumption within following hours. These criteria are refined according to a demand prediction model that dynamically previews and checks the consumption results. This proposed model creates an interface between all these systems. This interface is tested and evaluated according to a study of the São Paulo´s metropolitan area, “Sistema Alto Tietê". The efficiency of this proposed model is presented having reductions in the electric energy costs.

O presente trabalho propõe um modelo de operação sustentado por um sistema de suporte à decisão para operar a distribuição de água em tempo real atendendo a condições / restrições hidráulicas com o mínimo custo de energia elétrica. O atendimento às condições / restrições hidráulicas são avaliadas por um modelo simulador hidráulico previamente montado e calibrado. O conjunto de resultados avaliados pelo modelo de simulação hidráulica é analisado por um modelo de otimização proposto com solução de programação linear. As condições de operação em tempo real geram a necessidade de alimentação de informações operacionais automáticas a qualquer momento e com curto espaço de tempo – menor que horário. Para uma operação otimizada, previamente analisada por um modelo de simulação hidráulica cria uma condição critérios para uma previsão do consumo a ser atendido nas próximas horas. Um refinamento desses critérios são utilizados em um modelo de previsão de demanda de água que prevê e checa seus resultados de forma dinâmica. O modelo de operação proposto cria uma interface entre todos esses sistemas. Essa interface é testada e avaliada a partir de um estudo de caso aplicado no Sistema Adutor Metropolitano de São Paulo. A eficiência do modelo de operação proposto é apresentada tendo como resultado uma redução no custo de energia elétrica. / This assignment considers an operation model supported by decision support systems to operates the water supply systems in real time, considering the hydraulical conditions while achieving some performance goals, in this case, reducing electricity costs (minimization of pumping costs) the attempt of the hydraulic constraints are evaluated by an hydraulical simulator previously calibrated. The set of results are analyzed by an optimization model which uses a linear programming. The operation conditions in real time requires automatic feeding operational information shortly at any time (less than an hour) for an optimized operation, previously analyzed by a hydraulic simulation model with creates condition criteria of consumption within following hours. These criteria are refined according to a demand prediction model that dynamically previews and checks the consumption results. This proposed model creates an interface between all these systems. This interface is tested and evaluated according to a study of the São Paulo´s metropolitan area, “Sistema Alto Tietê”. The efficiency of this proposed model is presented having reductions in the electric energy costs.

The general aim of this research is to investigate approaches to: •improve small area market demand (i.e. SAMD) prediction accuracy for the purchase of automobiles at the level of each Census Collection District (i.e. CCD); and •enhance understanding of meso-level marketing phenomena (i.e. geographically aggregated phenomena) relating to SAMD. Given the importance of SAMD prediction, and the limitations posed by current methods, four research questions are addressed: •What are the key challenges in meso-level SAMD prediction? •What variables affect SAMD prediction? •What techniques can be used to improve SAMD prediction? •What is the value of integrating these techniques to improve SAMD prediction? To answer these questions, possible solutions from two broad areas are examined: spatial analysis and data mining. The research is divided into two main studies. In the first study, a seven-step modelling process is developed for SAMD prediction. Several sets of models are analysed to examine the modelling techniques’ effectiveness in improving the accuracy of SAMD prediction. The second study involves two cases to: 1) explore the integration of these techniques and their advantages in SAMD prediction; and 2) gain insights into spatial marketing issues. The case study of Peugeot in the Sydney metropolitan area shows that urbanisation and geo-marketing factors can have a more important role in SAMD prediction than socio-demographic factors. Furthermore, results show that modelling spatial effects is the most important aspect of this prediction exercise. The value of the integration of techniques is in compensating for the weaknesses of conventional techniques, and in providing complementary and supplementary information for meso-level marketing analyses. Substantively, significant spatial variation and continuous patterns are found with the influence of key studied variables. The substantive implications of these findings have a bearing on both academic and managerial understanding. Also, the innovative methods (e.g. the SAMD modelling process and the model cube based technique comparison) developed from this research make significant contributions to marketing research methodology.

The general aim of this research is to investigate approaches to: •improve small area market demand (i.e. SAMD) prediction accuracy for the purchase of automobiles at the level of each Census Collection District (i.e. CCD); and •enhance understanding of meso-level marketing phenomena (i.e. geographically aggregated phenomena) relating to SAMD. Given the importance of SAMD prediction, and the limitations posed by current methods, four research questions are addressed: •What are the key challenges in meso-level SAMD prediction? •What variables affect SAMD prediction? •What techniques can be used to improve SAMD prediction? •What is the value of integrating these techniques to improve SAMD prediction? To answer these questions, possible solutions from two broad areas are examined: spatial analysis and data mining. The research is divided into two main studies. In the first study, a seven-step modelling process is developed for SAMD prediction. Several sets of models are analysed to examine the modelling techniques’ effectiveness in improving the accuracy of SAMD prediction. The second study involves two cases to: 1) explore the integration of these techniques and their advantages in SAMD prediction; and 2) gain insights into spatial marketing issues. The case study of Peugeot in the Sydney metropolitan area shows that urbanisation and geo-marketing factors can have a more important role in SAMD prediction than socio-demographic factors. Furthermore, results show that modelling spatial effects is the most important aspect of this prediction exercise. The value of the integration of techniques is in compensating for the weaknesses of conventional techniques, and in providing complementary and supplementary information for meso-level marketing analyses. Substantively, significant spatial variation and continuous patterns are found with the influence of key studied variables. The substantive implications of these findings have a bearing on both academic and managerial understanding. Also, the innovative methods (e.g. the SAMD modelling process and the model cube based technique comparison) developed from this research make significant contributions to marketing research methodology.