Comparative Study of Prediction Models for Flight Departure Delays in Airports
This paper presents a comparative study on prediction models for flight delays in airports. Flight delay prediction is important to alleviate the cost occurred due to the same. The flight dataset has been collected from Kaggle for the O’Hare International Airport at Chicago for the year 2015.The prediction models explored in this paper include variations of gradient boosting machines, an important ensemble method for prediction analysis. The different boosters included are Tree, Linear and Dart. Each model is trained using the Caret Package in R to obtain results. Important parameters and the effect of tuning these parameters on the model prediction result are also discussed. xgbTree shows the best prediction results.
Keywords - Data Analytics, Flight Delay Predictions, Gradient Boosting.