Modelagem e previsão por meio de metodologia Box & Jenkins: uma ferramenta de gestão

• Ano de defesa: 2006
• Autor(a) principal: Kirchner, Rosane Maria
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Santa Maria; BR; Engenharia de Produção; UFSM; Programa de Pós-Graduação em Engenharia de Produção
• Palavras-chave em Português: Gestão; Séries temporais; Modelagem; Box-Jenkins; Management; Times series; Modeling; CNPQ::ENGENHARIAS::ENGENHARIA DE PRODUCAO
• Link de acesso: http://repositorio.ufsm.br/handle/1/8325

Independently of the company size it is necessary to establish goals, parameters and strategies. It is important to have guidance, a plan that conducts the actions for the present as a way to reach the goals for the future. Moreover, the planning of the company requires organization, controls, tools which assist helpful favor the decisionmaking and the necessary reconducts. The modeling and consequently the forecast are indispensable tools for the management, such, proves it contribution of the Box & Jenkins methodology. The market is highly competitive and selective, demanding not only efficiency, but mainly effectiveness. Actions directed to the magnifying of the quality of an organization are as important as the capacity to foresee, to know, to measure and to analyze the resolutivity of the used methodologies. In this context this present research is realized, searching to know the behavior of the company invoicing from an agricultural company and its two branch offices, molding and, in such a way, allowing putting into practice forecasts. Thus, the methodology applied consists of the use of historical data from the invoicing of the same ones. For the Main company, of Beta Company was found an ARMA(1,1) model, being this the most parsimonious. In CB Branch office the SARIMA (1,0,0) * (2,0,0) was the model that better described the data and in SL Branch office, the model SARIMA (1,0,0) * (1,0,0). Realizing the forecast and comparing with the real data placed by the company, it was verified the efficiency of the model, being that all the values meet up with the confidence interval of 95%. Taking into base the results gained in this research, it is possible to affirm that the methodology used constitutes of an important tool, being able to be used by the managers of the respective companies, supplying subsidies to the planning of the following months.