A STUDY ON PREDICTIVE MODELING OF FLIGHT DELAYS USING SUPERVISED LEARNING ALGORITHMS

Abstract

Air travel has been very successful since it is the fastest form of transportation, which has given people a lot of confidence in it over the years. Nevertheless, airlines have had to adjust to the issue of aircraft arrival delays, which arise from the fact that runway availability and airspace organization play significant roles in determining the time it takes for a flight to reach its destination. Accurately forecasting flight delays is crucial to the aviation industry's efficiency. In the center of recent research are machine learning strategies for forecasting aircraft delays. Algorithms for supervised machine learning have found widespread use in several areas of ML, including pattern recognition, data mining, and machine translation. Prediction methods in the past have often been limited to a single compass bearing or airport. Unpredictability is a major factor in flight planning, making it one of the most challenging conditions in business. Airports in Lithuania have had their flight time variance examined for this paper. SMOTE is employed to check the statistical reliability of the dataset. Have used the FCMIM method for feature selection. Predicting new delays in combat time using a supervised ML model is now possible. XG-Boost, LightGBM, and AdaBoost, three tree-boosting algorithms, were used in the study.