A0285
Title: Time series forecasting approaches to retail sales in EU Countries: Portugal and its major trading partners
Authors: A Manuela Goncalves - University of Minho (Portugal) [presenting]
Susana Lima - University of Minho (Portugal)
Marco Costa - University of Aveiro (Portugal)
Abstract: In the area of economics, particularly in the retail segment, sales forecasting supports most of the strategic planning decisions of any retail business. It must be as accurate as possible to ensure corporate profitability. The main goal is to evaluate forecasting methods' accuracy in the area of time series modelling applied to retail segment data. A method is proposed to compare the ARMA model's accuracy and their extensions, the classical decomposition time series associated with multiple linear regression models, and the exponential smoothing methods (Holt-Winters). These methods are chosen because of their ability to model trends and seasonal fluctuations present in economic data, particularly in retail sales data. The data available on the Eurostat platform correspond to monthly indexes of EU Countries' retail trade turnover. According to PORDATA, Portugal's major trading partners (import and export of goods and services) are Germany, Spain, France, Italy, the Netherlands, and the UK. The results suggest that multiple linear regression models are not the most appropriate to forecast retail sales indexes, while the SARIMA models are identified as the most accurate ones. Holt-Winters models are also a viable alternative, although they are not considered the most appropriate.