Flight Delays Prediction

Abstract

Flight Delay Prediction Remains A Significant Challenge In Modern Air Transportation Management. This Paper Presents A Comparative Analysis Of Machine Learning Approaches—including Classification-based, Ensemble-based, And Hybrid Predictive Models—for Forecasting Flight Delays. The System Integrates Data Preprocessing, Feature Selection, And Model Training Using Algorithms Such As Random Forest, K-Nearest Neighbors (KNN), Naive Bayes, And Logistic Regression. Key Flight Attributes Are Extracted From Historical Datasets And Refined Through Feature Engineering. Delay Prediction Is Further Enhanced By Applying Cluster Sampling Techniques For Balanced Data Representation. Experimental Evaluation Using U.S. Domestic Flight Data Revealed That The Random Forest-based Hybrid Approach Achieved The Highest Predictive Accuracy Of 89.3%, Slightly Outperforming Classification-only Models (KNN: 86.7%, Logistic Regression: 85.2%).

Keywords