Prediksi Keterlambatan Penerbangan Menggunakan Metode Decision Tree Untuk Penentuan Premi Asuransi Perjalanan

Abstract

Flight delays become an inevitable issue on flight commercial. Compensation regulated in the Ministerial Regulation considered disproportionate with occurring delays. Otherwise, airline company party are also reluctant to improve the quality of service, one of the reason is pay compensation cost for passenger is much less expensive than the cost for improving services. Therefore, a system needed for calculating the amount of premium that can be paid by passengers to benefit both parties. By using statistical calculation
method and machine learning algorithm, Decision Tree, delays can be predicted based on category of delays regulated in the Ministerial Regulation and insurance premium can be calculated accordingly and mutually beneficial to both parties. Phase of system design is as follows: read flight commercial in Indonesia from year 2017 to 2019 raw data, preprocess data, cleanse data, train data, process prediction, calculate premium and build visualization for presenting prediction result and premium price. Test result based on confusion matrix shows that model for predicting delays has an accuracy of 72.76%. Then from validation process, it obtained that similarity level of prediction result to validation result is 96.14%. The premium calculation result has premium value that is more reasonable and profitable for passenger flight.