Modelos de previsão da demanda para a Millagre Indústria de Alimentos Ltda

Abstract

Demand forecasting is a crucial tool for planning the productive process of companies. In this sense, the Department of Production Planning, Programming and Control is responsible for developing the ways in which the objectives of the organization supported in the demand forecasting will be achieved. It is no different when it comes to microenterprises, which need these same tools, demand forecasting and planning productive, to grow and remain competitive in the market. In many cases, small businesses do not have the necessary knowledge to develop and work with these tools in their day-to-day operations. In addition, this work sought to find a model of demand forecast for a microenterprise in the food industry of the city of Aracaju, Sergipe. From the catalog of items produced by the company, the products of class A were selected and included in the study. With the support of the SPSS, seven demand forecasting methods were modeled: Moving Average for three and six months, Linear Trend, Seasonal Simple, Seasonal Additive and Seasonal Multiplicative. In addition to these methods, the models were combined. As a measure of accuracy, the Mean Absolute Percentual Error (MAPE) indicator was used to verify the best fit models for the time series used. The data were adjusted by exploratory analysis using the software Studio R. For each item the two forecast models with the lowest MAPE were selected and the models were combined according to their forecasting average. The combination of the models was used to search for a better level of accuracy, because the time series presented points of randomness. The result obtained after the combination of the methods was not significant in relation to the models applied separately. The model that best conformed to the behavior of the analyzed time series was the Seasonal Additive, since it verifies a trend in the series with a seasonal component without the series amplitude changing with time. It is concluded that it is important to make forecasts with appropriate tools that have been developed for this purpose. It was also evidenced the need to record the sales history along with the indication of the actions that influenced the demand. In this way it will be possible to correctly interpret the data and then make forecasts with greater accuracy.