Airline Price prediction using ExtraTrees Regressor

Abstract

Airlines employ dynamic pricing strategies, which are based on demand estimation models, to set ticket prices. Airlines often set the price of their seats based on the necessary demand because they can only sell a certain number of seats on each journey. Airlines typically raise travel costs during times of strong demand, which slows down the rate at which seats are filled. When demand declines and seats become unsold, airlines typically lower their ticket prices to draw in new passengers. It will be more beneficial to sell those seats for more than the cost of service per passenger because unsold seats result in a loss of revenue. In order to forecast airline ticket prices, the primary objective of this study was to determine the factors influencing airline ticket costs and look into the relationships between them. Consequently, a model is created to predict the cost of plane tickets, allowing consumers to make better-informed decisions regarding their purchases. The cost of airline tickets in India is then predicted by this study using Four different machine learning algorithms: Extra Trees Regressor, XG Boost Regressor, Random Forest, and Decision Tree. Furthermore, hyperparameter optimization is done to obtain the most accurate and ideal prediction results. The Extrat Trees regressor yielded the greatest results, with the lowest RMSE of about 1807.59 and the highest accuracy of nearly 88%.