International Journal on Recent and Innovation Trends in Computing and Communication ISSN: 2321-8169 Volume: 6 Issue: 4 296 - 298 ______________________________________________________________________________________ 296 IJRITCC | April 2018, Available @ http://www.ijritcc.org _______________________________________________________________________________________ Optimizing Airline Ticket Purchase Timing Manan Dedhia Department of Computer Engineering K.J. Somaiya College of Engineering Mumbai-77, India manan.dedhia@somaiya.edu Rahul Jagdale Department of Computer Engineering K.J. Somaiya College of Engineering Mumbai-77, India rahul.jagdale@somaiya.edu Amit Jadhav Department of Computer Engineering K.J. Somaiya College of Engineering Mumbai-77, India amit.dj@somaiya.edu Prof. Bhakti Palkar Department of Computer Engineering K.J. Somaiya College of Engineering Mumbai-77, India bhaktiraul@somaiya.edu Abstract—Our approach in this paper is to suggest the user to either buy or wait for the purchase of airline tickets.Airline tickets prices are volatile and keep on varying depending on various parameters. Users, not having much information about these parameters, are often forced to buy tickets at high prices. This paper proposes a machine learning based prediction system which uses logistic regression to suggest users to buy the ticket, implying that prices are going to rise in coming days or wait for some time implying prices are going to plummet in coming days. This system also predicts the price of the date user wants to travel. Keywords-Machine learning; Logistic regression;Air fare; prediction; scrapping. __________________________________________________*****_________________________________________________ I. INTRODUCTION There is always a high demand of airline tickets and in absence of proper knowledge, users often do not have the luxury to book tickets at the best prices, usually ending up paying higher rate for the seat. This is further complicated by the confidential policies of airline companies, restricting the flow of information towards users which may be helpful to predict when to buy tickets. This void can be filled by our prediction system which will predict the prices of the tickets on the day that user wishes to travel and to suggest user whether to buy or wait for the ticket. II. PROPOSED APPROACH The proposed system has been divided into following modules. The first module was to choose an appropriate algorithm which would be the most efficient and accurate. Few algorithms that were considered for the purpose of this project were support vector machine, linear regression and logistic regression. As the objective of the system is to provide either of the two values (BUY or WAIT), logistic regression was the most suited algorithm for the system. A rudimentary model was created using the chosen algorithm which would predict the fare for any given day and inform user to either buy or wait. The second module was to scrap the data online from expedia.com website. For any prediction or classification problem, we need historical data to work with so as to run machine learning algorithms on it. For this system, we need to have comprehensive data of past flights on each of the routes considered. For this purpose, a python script was written which collected all the necessary data at a specific time daily. The script curated the following significant parameters from the website: 1. Arrival Airport 2. Arrival Time 3. Departure Airport 4. Departure Time 5. Plane type Name 6. Airline Name 7. Flight Duration 8. Plane Code 9. Ticket Price 10. Number of Stops The third module was to clean and prepare the data for further processing. Data needs to cleaned and prepared according to the model's requirements. This is the most important and time- consuming step for any machine learning model.Other task included creating a user interface which was easy enough for the user to understand and self-explanatory.