Evaluating the impact of climatic variability in wine production

• Autor(a) principal: Rui Miguel Cruz Soares Pinto
• Data de Publicação: 2016
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: https://repositorio-aberto.up.pt/handle/10216/88555
• topic: Ciências da engenharia e tecnologias; Engineering and technology

In the wine production business climate variability is one of the most critical conditions, being essential in regards to the process of ripening fruits so that it possesses the required characteristics to produce a good wine. Adding to this factor, the climatic variations have disastrous consequences not only for wine producers and workers but also for the land used for vineyards. Performing a good forecasting and statistical analysis of the wineries previous productions can help businesses save money and preserve the environment. In regards to this problem, new solutions arise for the processing of "datasets" based on data from winery production of past years and with further analysis, it is possible to achieve and identify the different climatic components and their impact on wine production. The solution would be based on the premise of "Machine Learning" consisting in building a model based on existing data provided by the"Dataset" [3] in order to be able to group similar data into subgroups according to its characteristics, making use of Decision trees, more specifically Regression trees to perform the grouping. This process would demonstrate the relationship between the meteorological series and its impact on winery production. The solution presented is an easy to understand way to represent grouping of data. The main goal of this thesis is to establish the impact of the different climacteric conditions (temperature, precipitation) in wine production. The solution in this dissertation is based on the premise of machine learning and aims to innovate in order to apply the concept of decision trees to multiple time series, aiming to identify subgroups of data and grouping them into classes.