A Flight fare Prediction Using Machine Learning

Abstract

While airlines (the sellers) always work to increase their revenue by changing pricing for the same service, air travellers (the buyers) frequently search for the ideal time of year to buy flights in order to save as much money as possible. The choice to raise or lower tickets at various points leading up to departure dates can be made by the sellers based on all the relevant data, such as historical sales, market demand, consumer profile, and behaviour. The buyers, on the other hand, have limited access to data to help them decide whether to delay or make a quick flight purchase. In this study, we suggest a new model that might assist the purchaser in anticipating price movements even in the absence of official airlines. Our results showed that the suggested model, despite lacking several essential components, such as the number of unsold seats on flights, can forecast trends as well as actual changes in airfare up to the departure dates using public airfare data that is readily available online. We also determined the characteristics that have the biggest effects on changes in airfare.

Introduction

I. INTRODUCTION

Due to the Internet's and e-phenomenal commerce's expansion, travellers may now readily check the prices and availability of all airlines worldwide. These consumers can purchase their chosen tickets online through official airline or agent websites once they are satisfied with an airfare. A number of prediction models have been developed to forecast the airfare before departure in order to assist clients in purchasing the least priced flights. To generate "purchase" or "wait" signals for clients, different data mining techniques as well as time series data analytics were applied [2]. Regression techniques such as partial least square regression [3] and linear quantile mixed regression are also used to build prediction models. The fact that the aforementioned prediction models exclusively concentrate on mature aviation markets, such those in the US, is a recurring theme. Additionally, a model that can forecast the airfare more accurately and ultimately help clients make better decisions to get the best airfare can be built using more of the airfare information that can be collected online. Currently, just a handful of these features are used to predict the price.Waiting to buy the cheapest airfare may result in missing the flight because it is impossible to predict when all of the seats on a given aircraft are sold out. It's also intriguing to learn which characteristics have the biggest effects on changes in airfare before departure dates.

II. RELATED WORK

1. K. Tziridis, Th. Kalampokas, G.A. Papakostas K.I. Diamantaras: The issue of predicting ticket rates is covered in this essay. In order to achieve this, a collection of characteristics that define a typical flight are chosen, presuming that these characteristics have an impact on the cost of an airline ticket. Eight cutting-edge machine learning (ML) models are employed to forecast the pricing of airline tickets using the attributes, and the models' performance is compared to one another. This study examines the relationship between forecast accuracy and the feature set used to represent an airline, in addition to the prediction accuracy of each model.
2. Tao Liu, Jian Cao Yudong Tan, Quanwu Xiao: In this study, we offer the ACER context-aware ensemble regression model, which incorporates various context-aware models and adaptively modifies context features. Context characteristics are arbitrarily chosen to efficiently cluster data, and several regression models are trained for data with various contexts. This approach is inspired by bagging and boosting. The context feature list is additionally constantly modified by removing some unnecessary elements. Our model is contrasted in the experiment on the real data set with the baseline regression model, random forest, and traditional time series models. The outcomes demonstrate that ACER outperforms the other models by a wide margin.
3. Viet Hoang Vu, Quang Tran Minh, Phu, H. Phung: In this article, we provide a brand-new model that might assist the customer in anticipating price trends without relying on official airline information. Our results showed that the suggested model, despite lacking several essential components, such as the number of unsold seats on flights, can forecast trends as well as actual changes in airfare up to the departure dates using public airfare data that is readily available online. We also determined the characteristics that have the biggest effects on changes in airfare.
4. William Groves, Maria Gini: However, only sporadically does the earliest buying method result in the ideal lowest cost ticket. This paper suggests a model for determining the best course of action for potential departures. The ultimate use of this concept is to automatically make daily purchases on behalf of purchasers of airline tickets in order to reduce their costs.
5. K.I. Diamantaras, T. Papadimitriou: In order to handle medium scale data in applications involving pattern classication, this study introduces a parallel version of the kernelized Slackmin method. The fundamental ideas of the serial Slackmin method are first discussed, with emphasis on how easily it may be parallelized due to its parallel nature. Utilizing the parallel processing features of a low-cost NVIDIA GPU card's CUDA architecture, parallelization is made possible.
6. Hang Zhou, Weicong Li, Ziqi Jiang, Fanger Cai and Yuting Xue: In order to identify the elements affecting flight operation, this study first describes the factors impacting flight operation in previous research findings. It next analyses and filters the factors. The GRU neural network model is then created and validated using actual flight data. Finally, the benefits of the model developed in this research are highlighted and contrasted with a number of widely used neural network models and random forest models in machine learning.
7. Micha Zoutendijk, Mihaela Mitici: In the present study, we derive probabilistic delay forecasts for flights landing and taking off from a local reference airport. To the best of our knowledge, this marks the first instance in which probabilistic projections for individual flight delays are made. We use the mixture density networks and random forest regression machine learning methods. We take into account features based on the flight schedules that are accessible at the reference airport as well as the weather data gathered at the airports where the flights originated and ended. The performance of the examined machine learning methods, which calculate delay probability density functions, is evaluated using appropriate metrics (pdf). Additionally, the effect of these algorithms' hyperparameter selection is examined.
8. O. Basturk, C. Cetek: In this paper, machine learning algorithms are given for predicting aircraft Estimated Time of Arrival (ETA). The management of delays and air traffic flow, runway and gate assignment, collaborative decision-making (CDM), coordination of ground personnel and equipment, and optimization of arrival sequence, among other things, depend on accurate ETA forecast. Machine learning can create predictions with flimsy or no assumptions while learning from past data.
9. Guan Gui, Fan Liu, Jinlong Sun, Jie Yang, Ziqi Zhou: This study analyses various machine learning-based models for generalised flight delay prediction tasks and investigates a wider range of variables that could potentially affect flight delays. Automatic Dependent Surveillance Broadcast (ADS-B) messages are received, pre-processed, and combined with additional data, such as weather, flight information, and airport information, to create a dataset for the proposed scheme.
10. Bin Yua, Zhen Guo, Sobhan Asian , Huaizhu Wang Gang Chen: This study examines high-dimensional data from Beijing International Airport and offers a useful model for forecasting aircraft delays. A novel deep belief network method is used to mine the internal patterns of flight delays after taking a multifactor approach. The created model incorporates support vector regression to carry out supervised fine-tuning inside the described predictive architecture.

III. EXISTING SYSTEM/OPEN ISSUES

In this article, we provide a brand-new model that might assist the customer in anticipating price trends without relying on official airline information. Our results showed that suggested model, despite lacking several essential components, such as the number of unsold seats on flights, can forecast trends as well as actual changes in airfare up to the departure dates using public airfare data that is readily available online. We also determined the characteristics that have the biggest effects on changes in airfare.

A. Negative Comments

The traditional airfare price prediction system is limited by the nonlinear interrelationship of multiple factors and fails to deal with the 

Conclusion

In this study, we evaluated a number of traditional machine learning algorithms on our airfare dataset to create a comprehensible prediction model that can forecast the trend of airfare in a developing aviation market (Vietnam) and assist customers in choosing the best time to book flights in order to maximise savings. Using our own technique, we used information that was gathered from internet travel agency websites. We could only obtain a small amount of publicly available data, missing important details like the number of unfilled seats on a flight. By layering two independent prediction models, Random Forest and Multilayer Perceptron, we may create a final interpretable prediction model. We stack data using fine-tuned weights, with R-squared serving as the primary evaluation metric..

References

[1] K. Tziridis, Th. Kalampokas, G.A. Papakostas K.I. Diamantaras, “Airfare Prices Prediction Using Machine Learning Techniques”, IEEE Commun. Mag., vol. 54, no. 5, pp. 138–145, May 2017. [2] Tao Liu, Jian Cao Yudong Tan, Quanwu Xiao , “ACER:An Adaptive Context-Aware Ensemble Regression Model for Airfare Price Prediction”, IEEE Commun.(2017). [3] Viet Hoang Vu, Quang Tran Minh, Phu, H. Phung, “Airfare Prediction Model for Developing Markets”, 2018. [4] William Groves , Maria Gini, “Optimal Airline Ticket Purchasing Using Automated User-Guided Feature Selection”, IEEE, 21-25 July, 2013. [5] K.I. Diamantaras, T. Papadimitriou, “Parallel pattern classification utilizing GPU-based Kernelized Slackmin algorithm”, IEEE, 2016. [6] Hang Zhou, Weicong Li, Ziqi Jiang, Fanger Cai and Yuting Xue, “Flight Departure Time Prediction Based on Deep Learning”,– IEEE, March , 2022. [7] Micha Zoutendijk, Mihaela Mitici, Probabilistic Flight Delay Predictions Using Machine Learning and Applications to the Flight-to-Gate Assignment Problem”, IEEE, JULY 2021. [8] O. Basturk, C. Cetek, “Prediction of aircraft estimated time of arrival using machine learning methods”, IEEE, 2021., [9] Guan Gui, Fan Liu, Jinlong Sun, Jie Yang, Ziqi Zhou, “Flight Delay Prediction Based on Aviation Big Data and Machine Learning”, IEEE,t 2019 [10] Bin Yua, Zhen Guo, Sobhan Asian , Huaizhu Wang Gang Chen, “Flight delay prediction for commercial air transport: A deep learning approach”, IEEE, 2019

Copyright

Copyright © 2022 Ketan Jayatkar, Dipak Jagtap, Pratik Dengale, Aditya Satam, Prof Mahendra Nivangune. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.