COSTA JUNIOR, Carlos Tavares da2025-03-242018PEREIRA, Olavo Gomes. Sistema Fuzzy para Avaliação de Desempenho de Programas de Governo. 2018. Dissertação (Mestrado Profissional) – Universidade Federal do Pará, Instituto de Tecnologia, Programa de Pós-Graduação em Engenharia de Processos, 2018.https://rigalileo.itegam.org.br/handle/123456789/830This work implemented a Fuzzy System of Evaluation of Government Programs, evaluation of the mid-term type, presenting improvements in relation to the current system, called Preventive Control System - Monitoring and Evaluation of Programs (SICONP-MAP), in the treatment of points of singularity and cloudiness, with potential possibility of incompleteness treatment. The modeling was implemented in the JuzzyOnline and RProject platforms, where both evaluated the "Security for Peace" Program and presented 99.99% similarity, 0.101914 for the maximum difference of the residual values, 0.003688 for the median, and 0.035 for the standard deviation. These metrics showed that there were no differences between the implementations with a level of significance lower than 0.1%, where it is inferred that the modeling is robust. Robustness was compared with SICONP-MAP, where the Fuzzy System presented an accuracy of 88.88%, corresponding to 24 of 27 actions evaluated according to reality, against 44.44% of SICONP's accuracy associated with 12 actions with assessments adhering to reality. Considering that the Fuzzy System points out the distribution of stratified actions in the field of evaluation, i.e., the performance characteristic of the Program based on actions, it was observed the distribution of the expenditure realized in the field of evaluation, and then it was verified to what extent the first distribution explains the distribution of the expense. The sequences presented 94.82% similarity, 0.029291 for the maximum difference of the residual values, 0.008312 for the median, and 0.007549 of standard deviation, confirming a significance lower than 0.1%. Despite the high correlation, the metric of the expense, which points to the quality of the expenditure, evidenced the risk of classifying 43.64% of the expenditure as high performance whose actions were included as suspected of "Planning Failure". The implementation can be extended to all spheres of government, applying directly to any Program that does not present incompleteness, otherwise, it is necessary to pre-treat the data.pdf.Engenharia de ProduçãoSimulação por ComputadorPolítica PúblicaAvaliação de ProgramasSistemas DifusosDissertaçãoEngenharia de Produção