Improving Airport Flight Prediction System Based on Optimized Regression Vector Machine Algorithm
Nowadays, the safest and most extensive type of transportation is air flights, and tens of thousands of flights are carried out from different airports every day around the world. However, many of these flights have short-term or long-term delays, which cause the passengers' plans to change, a...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
middle technical university
2024-09-01
|
| Series: | Journal of Techniques |
| Subjects: | |
| Online Access: | https://journal.mtu.edu.iq/index.php/MTU/article/view/2481 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Nowadays, the safest and most extensive type of transportation is air flights, and tens of thousands of flights are carried out from different airports every day around the world. However, many of these flights have short-term or long-term delays, which cause the passengers' plans to change, and as a result, the airlines suffer losses. the first step to deal with these air delays is to predict them air delay. Regression functions and machine learning algorithms have a very good performance in predicting time series data. In this research, the optimized support vector regression (SVR) algorithm has been used to improve the accuracy of air delay prediction. The SVR algorithm is a machine learning algorithm that uses regression functions. The SVR algorithm has hyperparameters that play a key role in its performance and setting these parameters is very important. In this research, to improve the performance of the SVR algorithm in predicting air delays, the particle swarm optimization (PSO) algorithm has been used to adjust the hyperparameters of the SVR algorithm. The database used in this research was taken from the US Air Transport Fleet website and is an unbalanced database. Based on the results obtained from the simulations, our proposed method has improved with an average accuracy of 87.41% compared to other compared works.
|
|---|---|
| ISSN: | 1818-653X 2708-8383 |