| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 3.09 MB | Adobe PDF |
Orientador(es)
Resumo(s)
This work project aims at supporting managers in explaining and predicting sales of specificproducts based on a devised methodology. Product sales time series were analysed andprocessed in order to select the best model type: explanatory models (through ordinary leastsquares method), univariate models (Box Jenkins methodology) or dynamic models mixing upthe two previous approaches. An automatic procedure to put the methodology in practice wasimplemented using Python, due to the huge amount of product sales to be modelled. Theprocess was tested using data from more than 1500 products from Beiersdorf Lisbon. For thesake of confidentiality, the names of the products were modified. The most accurate models aredescribed and analyzed.
Descrição
Palavras-chave
Regression Arima Box Jenkins Sales model
