Construcción de modelos estadísticos basados en serie temporales históricas del cultivo de yuca en Brasil

摘要

Cassava (Manihot esculenta Crantz) is a plant native of America and it has great importance atboth, worldwide and Brazilian. It is being commercialized both in natural form and in the flour and starch forms for the domestic market. It is exported and used in the most diverse industrial sectors such as food, pulp and paper, knitting, paints, glues, medicines, among others. Evaluating and interpreting historical data is fundamentally important for decision making, investment orientation and to designf public policies for agribusiness. The objective of this research was to study the context of cassava cultivation and the application of statistical forecasting tools based on time series analysis of historical data. Data obtained directly from the CEPEA and IBGE databases were used for the construction of the models in the year 2015 and later for validation of the results and realization of new forecasts in 2017. Main results verified that both, the regression analysis and the prediction by exponential smoothing methods Holt Winters forecast and the Box Jenkins method, were efficient in forecasting the time series studied. It alsoshowed that Brazil has been presenting a linear reduction in the area destined to cassava cultivation. The States that have contributed the mostto this area reduction are Maranh?o and Bahia. On the other hand, the States of Pará and Paraná stood out as the largest producers of cassava. The first one stands out in the production of cassava flour, while the second one in the production of starch. Finally, it was verified with the present study that the reduction of area diagnosed over the years is worrisome for the agroindustries of the sector, which require the cassava as raw material.