Unravel Before You Travel-A Decision System for Air-ticket Purchase

Unravel Before You Travel-A Decision System for Air-ticket Purchase

Abstract

Given past two years of airline ticket price data between two cities, we had to decide whether to recommend a customer to ‘Buy’ or ‘Not to Buy’ an airline ticket on date X (say, today) if he wanted to travel on date Y in future (say, 5 days from today) such that the purchase is profitable. If the air ticket prices for the next 5 days are known, we can decide if it will be profitable to buy the tickets today (if ticket price increases in future) or is it profitable to buy the tickets within the next 5 days (if ticket price drops in future).

Autoregressive integrated moving average (ARIMA), dynamic linear modeling (DLM) and double exponential moving average (DEMA) were used to predict the prices or trend in future for making decisions. Performance of the models is evaluated and compared based on the number of correct decisions made (we had the ground truth).Using ARIMA and DLM, future prices are predicted and the decision making is based on whether the predicted prices are higher or lower than today and it is also predicted that on which day will it be most profitable. Using DEMA, the trend is analyzed to make the decision only for the present day and no prediction is made about the future.


Parneet Kaur

Written by

Postdoc Scientist at Johnson & Johnson *Opinions are my own*

    Updated