A Comparative Analysis for Predicting Airline Arrival Delays

Abstract

Flight data is a large source of big data. Million flights are delayed or canceled each year due to several factors. Study aviation systems are being significant to the economy which improves customer satisfaction, and saves time. Delay Prediction in aviation systems is somewhat complicated because of the large volume of data, the multiple causes of delays. The reasons vary from region to region and from company to another. In this paper, we compare the performance of different machine learning approaches (Random Forest Classifier, logistic regression, Gaussian Naive Bayes and Decision Tree Classifier) for predicting the arrival delay depending on the multiple characteristics and mention the features in each approach. Using machine-learning toolkit supported on the Splunk platform to make a comparison between them. The Airline On-Time Performance Data are used for evaluating the models. The results demonstrate that the Logistic regression is better than others and works well with discrete data.