Aplicação de técnicas de previsão de demanda em uma loja de departamento

Abstract

In highly competitive environments, particularly in the retail sector where profit margins are narrow, the study of demand forecasting techniques assumes significant importance. Such techniques assist organizations in making informed decisions regarding product availability and distribution, thereby enabling effective planning and contributing to improved outcomes. In this context, the present study aimed to apply various demand forecasting techniques to historical sales data of portable home appliances from a department store chain operating in the cosmetics sector. The research examines the feasibility of applying these techniques and compares the results obtained. To this end, the ABC Curve was utilized to classify and select the most relevant products for analysis, accompanied by a preliminary assessment of the behavior of the historical sales data for the selected items. Based on this analysis, the following techniques were applied: Moving Average, Simple Exponential Smoothing, Exponential Smoothing with Trend and Seasonality and SARIMA. After generating the forecasts, evaluation metrics were employed to compare and assess the effectiveness of each forecasting method. This study concludes that the application of demand forecasting methods based on historical sales data is feasible and provides a comparative analysis of the performance across different selected products. Furthermore, forecast errors were analyzed by comparing predictions to the most recent actual results obtained by the company under study. Lastly, the behavior of the data was discussed, highlighting that certain products may exhibit more regular patterns in terms of trend and/or seasonality, while others may present irregular or random patterns, which can significantly influence forecasting accuracy.