| Abstrakt: |
This study aimed to find the optimal price for an airline ticket while ensuring that the customer does not cancel their booking. A dataset provided by an online airline reservation company in Pakistan named SASTAticket was used, which consisted of approximately 1 million records with 12 features. The independent variables included the booking date, booking time, departure date, departure time, refundable ticket, baggage weight, and baggage pieces, while the output variable was the target price to be predicted. Correlation analysis revealed a weak but significant relationship between the target variable and the baggage weight, baggage pieces, and booking month. Three machine learning algorithms were implemented, namely Decision Tree, Random Forest, and Gradient Boosting Regression, to predict the optimal price. The models were trained and tested using a 70/30 train-test split, and the results showed that Gradient Boosting Regression outperformed the other two models, with an R2 score of 0.89 and RMSE of 1092. The study's findings suggest that machine learning models can effectively predict optimal prices for airline tickets. The results of this research can help airlines to offer customers competitive prices while ensuring that they do not cancel their bookings, resulting in increased revenue and customer satisfaction. [ABSTRACT FROM AUTHOR] |
