Choose Language Hide Translation Bar

Predicting New User's Country Destination in Airbnb (2019-US-EPO-253)

Level: Intermediate


Aparna Gopakumar, Student, University of Connecticut School of Business


Airbnb, established in 2008, is an online hospitality company headquartered in San Francisco, CA. The company focuses on a hospitality service where users can act as hosts, often hosting their homes or locations to guests, who schedule and rent out locations to live as a subletting system. This is primarily done through a desktop website and mobile app, and the service has nearly 150 million guests with hosts in 81,000 cities globally. Focusing on the new user booking experience is of utmost importance to them to convert those new users into loyal customers. A new user can book a place to stay in more than 34,000 cities across more than 190 countries. By predicting the destination a new user may book, Airbnb can share more personalized content, decrease the average time to first booking and better forecast demand. Manipulating a few principal features of JMP, a gradient boosting model was built to predict the country in which a new user might book his destination. In the process, JMP was also used for the purpose of data exploration, data preprocessing and data visualization.