Machine Learning Approach of Predicting Airline Flight Delay using Naïve Bayes Algorithm

The aviation industry plays a critical role in global transportation, facilitating economic growth and revolutionizing travel. However, flight delays have become a growing concern, impacting both airlines and passengers. This study aims to study the Naïve Bayes algorithm for flight delay prediction. The objective is to develop a reliable flight delay prediction model using the Naïve Bayes algorithm and evaluate its performance. The data set that records flight delay and cancellation data from U.S Department of Transportation’s (DOT) was used for the prediction. This study has modified the parameter tuning for Gaussian Naïve Bayes to identify optimum values specifically to construct model for this flight delay dataset. The performance of parameters tuning Gaussian Naïve Bayes model was compared with another two well-known algorithms which are K-Nearest Neighbors (KNN) and Support Vector Machine (SVM)). The KNN and SVM algorithms were also trained and tested to complete the binary classification of flight delays for benchmarking purposes. The evaluation of algorithms was fulfilled by comparing the values of accuracy, specificity and ROC AUC score. The comparative analysis showed that the Gaussian Naïve Bayes has the best performance with an accuracy of 93% and KNN has the worst performance with ROC AUC score 63%.