Estudo de atraso de voos no aeroporto de Uberlândia via regressão logística

Abstract

The current work aims to evaluate, using logistic regression, which factors have influence on the flights delay at Uberlândia airport, concerning a certain airline. The datas were collected on the aircraft arrival and departure administrative sector of an airline that lands at the airport. More than 3,121 flights were analyzed from May 2017 to April 2019. 8 variables were analyzed, which are possible reasons for aircraft time on the ground that can cause flight delays. It was observed, among the studied variables, there were 4 really significant, concerning the response variable. The Stepwise criterion with the Akaike statistic was used to select the most parsimonious model and to measure the quality of the model fit. The classification matrix, the ROC Curve, was applied to organize, visualize and classify the model based on its predictive performance and Pseudo R² of Nagelkerke.