The planning and control of supply chain activities depend on accurate estimates of the amount of products and services to be processed, and estimates come as forecasts. One of the biggest challenges for food and beverage manufacturers is to adjust production and stocks to reduce product losses due to short-term perishability. Time series analysis is very important in a wide range of applications, especially when it comes to prediction and encompasses many different prediction models. However, it is necessary to determine which model best suits each situation. There are countless forecasting models presented in the bibliography. From market research to the most complex calculation methods. The reference shows three main groups of these methods, namely: qualitative, historical projections, and causal. Depending on the series and the desired prediction time, it is possible to choose the technique that works best. In addition to selecting the best technique, the prediction that the selected model will generate should be as close to the actual as possible. In other words, forecasting errors need to be minimized, so production managers plan market negligence and minimize costs. Demand forecasting methods can be based on mathematical models that use historical data or on qualitative methods, planned according to administrative experience and customer reviews.