Técnicas de previsão de demanda aplicadas em uma microempresa de panificação da cidade de Jaguapitã, PR

Abstract

This study aims to develop demand forecasting for products of a bakery using different statistical models, focusing on the application of ARIMA (AutoRegressive Integrated Moving Average) and linear regression. The research was conducted based on historical sales data of four products: French Bread, Panettone, Carolina Cookie, and Square Toast. The analysis of the linear regression models revealed that this approach was ineffective, presenting unsatisfactory results such as negative R² and high forecasting errors, especially during periods of high demand, such as weekends and holidays. In contrast, the ARIMA model proved to be more suitable for demand forecasting, providing more accurate predictions with lower errors. However, some limitations were observed, particularly regarding the predictability of demand for the Square Toast, which did not present a stationary behavior, complicating the forecasts. To improve the accuracy of the forecasts, it is suggested to increase the amount of data collected over time, which would allow for a more precise adjustment of the ARIMA model, resulting in a more assertive demand forecast. The continuous improvement of the ARIMA model could contribute to more efficient inventory management and more accurate production planning, benefiting the bakery in the long run.