Identificação de outliers em processos de dispensas e inexigibilidades em licitações públicas: um estudo comparativo entre UFRN, IFRN e UFERSA nos anos de 2017 e 2018

Abstract

The main function of the auditorship branch is the identification of anomalies. Given the large amount of data currently stored, the usual sampling-based procedures can detect only part of the problems. In view of this, data mining studies emerge, among which the quartile technique is highlighted. The present work aims to identify outliers through the study of data mining using the quartile method, in the processes of dismissal and unenforceability of bids performed by the Federal University of Rio Grande do Norte, Semi-Arid Rural Federal University and the Federal Institute of Education, Science and Technology of Rio Grande do Norte. To perform the analysis a quantitative approach was applied from the data collection on the three institutions studied. The information was extracted from the federal government transparency portal, between 2017 and 2018. By applying the quartile method it was possible to identify the values considered as outliers and the variations between light and extreme, according to the discrepancy. It was possible to observe the values and statistics related to outliers, as well as to determine the usefulness of the quartile method in identifying anomalies within the studied databases.