Modelo de previsão de vagas de educação infantil de um município brasileiro

Abstract

Education, as the right of everyone and the state duty and family, is a legal asset guaranteed by the Federal Constitution. In Brazil it is estimated that there are about 2.000.000 children deprived of access to early childhood education. A model capable of estimating enrollments is proposed to assist the formulation of strategies focused at attending the demand for early childhood education in a Brazilian municipality. Analyzing the budget for reforms and expansion of early childhood education and the number of enrollments, annually, it becomes clear that there is a mismatch between the amount invested and the increase in service?s capacity.Such behavior might justify itself by the unknown, albeit marginally, enrollments demand. Considering the ability to generalize and adapt of neural networks, the multilayer perceptron is used as a prediction model. Using enrollments historical series of Camboriú, between 2012 and 2019, trend and seasonality patterns were evaluated. Iteratively, neural network?s parameters were adjusted using the back propagation algorithm and the number of enrollments for the year 2019 was estimated, resulting in a average absolute percentage error of 1.44%. Even so, considering that it was used only a enrollments? series from one municipality, it is believed that using other variables and / or other architecture of neural networks could be further investigated. [resumo fornecido pelo autor]