Accurate forecasting is a real problem at many companies and that includes Alfa Laval in Lund. Alfa Laval experiences problems forecasting for future raw material demand. Management is aware that the forecasting methods used today can be improved or replaced by others. A change could lead to better forecasting accuracy and lower errors which means less inventory, shorter cycle times and better customer service at lower costs. The purpose of this study is to analyze Alfa Laval’s current forecasting models for demand of raw material used for pressed plates, and then determine if other models are better suited for taking into consideration trends and seasonal variation.

This thesis analyzes the application of functional link-net (FLN) method in forecasting electricity demand in Turkey. Current official forecasting model (MAED), which is employed by Turkish Electricity Transmission Company (TEiAS) and other methods are discussed. An emprical investigation and evaluation of using functional link nets is provided.

The economic evaluation of a new airport investment requires the use of estimated future air passenger demand.Today it is well known that air passenger demand is basicly dependent on various socioeconomic factors of the country and the region where the planned airport would serve. This study is focused on estimating the future air passenger demand for planned airports in Turkey where the historical air passsenger data is not available.For these purposses, neural networks and multi-linear regression were used to develop forecasting models. As independent variables,twelve socioeconomic parameters are found to be significant and used in models. The available data for the selected indicators are statistically analysed and it is observed that most of the data is highly volatile, heteroscedastic and show no definite patterns. In order to develop more reliable models, various methods like data transformation, outlier elimination and categorization are applied to the data.Only seven of total twelve indicators are used as the most significant in the regression model whereas in neural network approach the best model is achieved when all the twelve indicators are included. Both models can be used to predict air passenger demand for any future year for Or-Gi and Zafer Airports and future air passenger demand for similar airports. Regression and neural models are tested by using various statistical test methods and it is found that neural network model is superior to regression model for the data used in this study.

<p>Accurate forecasting is a real problem at many companies and that includes Alfa Laval in Lund. Alfa Laval experiences problems forecasting for future raw material demand. Management is aware that the forecasting methods used today can be improved or replaced by others. A change could lead to better forecasting accuracy and lower errors which means less inventory, shorter cycle times and better customer service at lower costs.</p><p>The purpose of this study is to analyze Alfa Laval’s current forecasting models for demand of raw material used for pressed plates, and then determine if other models are better suited for taking into consideration trends and seasonal variation.</p>

No abstract available. / Thesis (M.Ed.)-University of Durban-Westville, 1975.

The paper is concerned with the impact of market research prior to integration, on the structures of noncompetitive industries in integrated economy. The analysis focuses on separated, single commodity, monopolistic markets with stochastic demand. Monopolistic firms are considered in dynamic multiperiod model, where intertemporal links are determined by expenditures on market research in a present period and benefits from this activity (i.e., smaller variance of the prediction error) in the future. Assuming that each firm maximizes its total discounted expected utility from profit in indefinite time, we show that the optimal market research strategy is stationary and depends on market size. Consequently, in the period following integration firms operating prior to integration in small markets (such as Slovenia, Czech Republic, Hungary or Estonia) are expected to have much less information about the integrated market than their competitors operating before integration on European market. This informational asymmetry may affect the structure of the industry in integrated economy. In the extreme case, the firm operating before integration in the small market can be ruled out from the integrated market. (authors' abstract)

<p>This thesis discusses novel developments in Bayesian analytics for high-dimensional multivariate time series. The focus is on the class of multiregression dynamic models (MDMs), which can be decomposed into sets of univariate models processed in parallel yet coupled for forecasting and decision making. Parallel processing greatly speeds up the computations and vastly expands the range of time series to which the analysis can be applied. </p><p>I begin by defining a new sparse representation of the dependence between the components of a multivariate time series. Using this representation, innovations involve sparse dynamic dependence networks, idiosyncrasies in time-varying auto-regressive lag structures, and flexibility of discounting methods for stochastic volatilities.</p><p>For exploration of the model space, I define a variant of the Shotgun Stochastic Search (SSS) algorithm. Under the parallelizable framework, this new SSS algorithm allows the stochastic search to move in each dimension simultaneously at each iteration, and thus it moves much faster to high probability regions of model space than does traditional SSS. </p><p>For the assessment of model uncertainty in MDMs, I propose an innovative method that converts model uncertainties from the multivariate context to the univariate context using Bayesian Model Averaging and power discounting techniques. I show that this approach can succeed in effectively capturing time-varying model uncertainties on various model parameters, while also identifying practically superior predictive and lucrative models in financial studies. </p><p>Finally I introduce common state coupled DLMs/MDMs (CSCDLMs/CSCMDMs), a new class of models for multivariate time series. These models are related to the established class of dynamic linear models, but include both common and series-specific state vectors and incorporate multivariate stochastic volatility. Bayesian analytics are developed including sequential updating, using a novel forward-filtering-backward-sampling scheme. Online and analytic learning of observation variances is achieved by an approximation method using variance discounting. This method results in faster computation for sequential step-ahead forecasting than MCMC, satisfying the requirement of speed for real-world applications. </p><p>A motivating example is the problem of short-term prediction of electricity demand in a "Smart Grid" scenario. Previous models do not enable either time-varying, correlated structure or online learning of the covariance structure of the state and observational evolution noise vectors. I address these issues by using a CSCMDM and applying a variance discounting method for learning correlation structure. Experimental results on a real data set, including comparisons with previous models, validate the effectiveness of the new framework.</p> / Dissertation

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / The goal of the electricity distribution companies is to provide consumers with a continuous supply of energy and quality. This dissertation addresses the Vehicle Routing Problem, specifically the partially dynamic routing with static entries, where some events that occur stochastically are dynamically incorporated during the execution of the service. In this sense, we sought to develop a methodology to provide the emergency service events that arise randomly during the working day, taking into account attributes of location, time of service and time of occurrence, to minimize the travel time of vehicles on scheduled routes. For that, a sequence of steps has been developed and described for the structuring of a demand forecasting system, which should be able to design patterns and trends analyzed data from past demands. Intending to meet these assumptions, the study sought support in two forecasting methods: exponential smoothing and prediction from conditional probabilities. The study also sought to identify the main variables that influence the way aleatótia the occurrence of emergency orders. The results obtained with these methods, assisted in the capture of the stochasticity of the order process emergency orders, as well as in forecasting service demand. The work seeks to identify the input variables for routing, providing subsidies for the analyzed company that does not have this information. / A meta das empresas de distribuição de energia elétrica é proporcionar ao consumidor um fornecimento de energia contínuo e com qualidade. Esta dissertação aborda o Problema do Roteamento de Veículos, mais especificamente o roteamento parcialmente dinâmico, com entradas estáticas, onde alguns eventos que ocorrem de forma estocástica são incorporados dinamicamente durante a execução do serviço. Neste sentido, buscou-se elaborar uma metodologia capaz de prever as ocorrências de serviços emergenciais, que surgem aleatoriamente durante a jornada de trabalho, levando em consideração atributos de localização, tempo de serviço e horário de ocorrência, visando minimizar o tempo de deslocamento dos veículos nas rotas programadas. Para isso, foi desenvolvida e descrita uma sequência de etapas para estruturação de um sistema de previsão de demanda, o qual deve ser capaz de projetar padrões e tendências dos dados analisados a partir de demandas passadas. Pretendendo atender a estes pressupostos, o estudo buscou suporte em dois métodos de previsão: suavização exponencial e previsão a partir de probabilidades condicionais. O estudo ainda buscou, identificar as principais variáveis que influenciam de maneira aleatótia a ocorrência de ordens emergenciais. Os resultados obtidos com estes métodos, auxiliaram na captura da estocasticidade do processo de despacho de ordens emergências, bem como, na previsão de demanda de serviço. O trabalho busca identificar as variáveis de entrada para o roteamento, proporcionando subsídios para a empresa analisada que não dispõe destas informações.

Submitted by MARIO AUGUSTO MATOS SIMON ESTEVES (marioaugustoesteves@gmail.com) on 2016-12-21T01:17:05Z No. of bitstreams: 1 Dissertacao de Mestrado - Mario Esteves - Final.pdf: 3642994 bytes, checksum: 14617df83472dcb04e1f218abfd26cd4 (MD5) / Approved for entry into archive by Felipe Augusto Arakaki (arakaki@reitoria.unesp.br) on 2016-12-22T12:43:08Z (GMT) No. of bitstreams: 1 esteves_mams_me_guara.pdf: 3642994 bytes, checksum: 14617df83472dcb04e1f218abfd26cd4 (MD5) / Made available in DSpace on 2016-12-22T12:43:08Z (GMT). No. of bitstreams: 1 esteves_mams_me_guara.pdf: 3642994 bytes, checksum: 14617df83472dcb04e1f218abfd26cd4 (MD5) Previous issue date: 2016-10-21 / Com o contínuo crescimento da competitividade global, o grande desafio é trabalhar de forma enxuta, mas sem prejudicar o nível de serviço ao cliente. Para isso, busca-se uma rápida e adequada integração das necessidades do mercado na direção dos fornecedores, de modo a balancear e alinhar estrategicamente a demanda com a capacidade operacional ao longo de toda a cadeia de suprimentos. Para a presente pesquisa, utilizou-se levantamento do tipo survey, e o objetivo geral é verificar o panorama atual das práticas de Gestão de Demanda e Previsão de Demanda nas indústrias da Cadeia de Suprimentos Automotiva Brasileira, identificando as principais práticas utilizadas e as principais dificuldades relacionadas à execução dos processos de gestão e previsão de demanda, bem como as consequências causadas pelas variações e incertezas de demanda. Para tanto, com base na revisão da literatura e no método hipotético dedutivo de Popper, foi elaborado um questionário que foi respondido por 37 empresas da cadeia de suprimento automotiva dos mais diversos setores. Os resultados mostram que as empresas da cadeira de suprimento automotiva fazem uso com predominância de técnicas mais simples como opiniões de executivos e da equipe de vendas e utilização de médias móveis. A falta de disponibilidade de dados, a necessidade de capacitação e treinamento da equipe e a deficiencia no conhecimento dos modelos e ferramentas de previsão de demanda aparecem como as maiores barreiras para elaboração das previsões de demanda. / With the continued growth of global competitiveness, the challenge is to work lean way, but without affecting the level of customer service. As a result, a quick and proper integration of the market requirements towards suppliers should be sought, in order to balance and strategically align the demand with the operational capacity along the entire supply chain. This research use the survey method and the overall objective is to find what the current situation of Demand Management and Demand Forecasting practices in the industries of Brazilian Automotive Supply Chain, identifying the main practices and the difficulties related to the implementation of the management and demand forecasting processes, as well as those caused consequences as a result of variations and demand uncertainties. Therefore, based on the literature review and popper´s hypothetico-deductive method, it has been designed a questionnaire that was answered by 37 companies in the automotive supply chain in various sectors. The results show that companies in the automotive supply chair make use predominantly of the simplest techniques as executive and sales force opinion methods and use of moving averages. The lack of availability of data, the need of professional training and deficiency of knowledge of the models and demand forecasting tools appear as major barriers to development of demand forecasts.

Submitted by Geandra Rodrigues (geandrar@gmail.com) on 2018-01-10T17:04:37Z No. of bitstreams: 1 raphaelfranciscofirmianoteixeira.pdf: 2043800 bytes, checksum: c1621a5e2f6c91e329bbc4cd21d1701b (MD5) / Approved for entry into archive by Adriana Oliveira (adriana.oliveira@ufjf.edu.br) on 2018-01-23T11:53:03Z (GMT) No. of bitstreams: 1 raphaelfranciscofirmianoteixeira.pdf: 2043800 bytes, checksum: c1621a5e2f6c91e329bbc4cd21d1701b (MD5) / Made available in DSpace on 2018-01-23T11:53:03Z (GMT). No. of bitstreams: 1 raphaelfranciscofirmianoteixeira.pdf: 2043800 bytes, checksum: c1621a5e2f6c91e329bbc4cd21d1701b (MD5) Previous issue date: 2017-08-30 / Frequentemente os valores dos parâmetros exigidos em um Contrato de Fornecimento de Energia Elétrica para consumidores não industriais são estimados com base na previsão de demanda utilizando o método “Naive”, por vezes, com algum ajuste empírico, o que pode gerar um contrato não-ótimo para o consumidor. Exemplo desse tipo de consumidor são as universidades, principalmente as públicas, por possuírem dimensões físicas consideráveis. Em consumidores com esse tipo de comportamento, a elaboração de um perfil de demanda baseado em estudo do funcionamento das instalações torna-se algo muito complicado. Tendo em vista tratar-se de um consumidor pertencente ao Serviço Público, há a necessidade de Contratos definidos com critérios suficientemente claros, haja vista a pressão dos órgãos de controle. Mais ainda quando se considera o uso responsável e eficaz do dinheiro público. Portanto, métodos com base na previsão de demanda do consumidor, em função do seu histórico e capazes de uma aproximação maior com a realidade, seriam importantes para obter contratos com valores financeiros minimizados. Tendo os dados de Demandas Registradas da Universidade Federal de Juiz de Fora e dados auxiliares de Temperaturas e Calendário de Aulas, desenvolvemos um método que testa previsões realizadas por métodos lineares (Médias Móveis, ARIMA e Holt-Winters), com previsões realizadas por métodos não- lineares (Redes Neurais). Comparamos estas previsões, e a melhor foi levada a um processo de otimização utilizando Algoritmos Genéticos. Essa otimização revelou dados ótimos para o Contrato e os respectivos custos. A previsão com melhor desempenho foi a obtida utilizando-se Redes Neurais, sem os dados auxiliares. A otimização levou a escolha da Tarifa Azul, com previsão de ganhos econômicos para a UFJF. / The values of the parameters required in a Contract of Electric Power Supply are often estimated with bases on the demand forecasted using the “Naive” method, for nonindustrial consumers. Sometimes, the method suffers some empirical adjustment, which can generate a non-optimal contract for the consumer. Universities (in special the public ones) are examples of these type of consumers since they have considerable physical dimensions. The elaboration of a demand profile for these type of consumers, based on a study of the operation of the facilities, is a complicated task. Because the consumer is part of the Public Service, there is a need for Contracts defined with sufficiently clear criteria, given the pressure of the control bodies, particularly when the responsible and effective use of the public money is considered. Therefore, methods based on the consumer demand forecast in function of consumer’s history, and capable of greater approximation with reality, would be important to obtain contracts with minimized financial values. A method was developed based on data of the registered demands of the Federal University of Juiz de Fora (UFJF), and Temperature and Class Calendar adjuvant data. The method tests the predictions made by linear (Moving Averages, ARIMA and Holt-Winters) and by non-linear methods (Neural Networks). The predictions were compared and the best one was taken to an optimization process using Genetic Algorithms. The optimization revealed optimal data for the contract and its costs. The prediction that showed the best performance was the one obtained using Neural Networks without the adjuvant data. The optimization led to the choice of the “Tarifa Azul”, with possible economic gains for UFJF.