Previsão do custo de manutenção predial em universidades públicas federais por modelagem com regressão linear - RLM

Abstract

The present work consists of the development and evaluation of a tool that helps in predicting the cost of building maintenance of an IFES This forecast, besides to contributing to the planning of maintenance services, assists IFES to comply with specific legislation that deals with budgetary planning of institutions and the Union. To ensure that the prediction of the cost of building maintenance has robustness, mathematical model were developed, using techniques with RLM that perform the prediction of the values of these costs, aiming to help IFES in carrying out a more efficient financial planning for performing maintenance tasks. As an object of study, was used the set of 63 Federal Universities in Brazil that were in operation in 2018 . Based on annual data from these IFES, information that influences the cost of building maintenance in the following year is analyzed. The analysis of the model’s statistical parameters showed robustness in meeting the applied statistical assumptions, presenting indices R = 90, 4% and R2 = 81, 8%. A set of simulations with the final equation reveal satisfactory results for the developed tool.