Uncovering Clusters of Airbnb listings in Tasmania, Australia

This is one of our mini-projects for our Data Mining and Wrangling class in collaboration with Jasper Pangan and Cedric Corro.

Summary

Australia’s largest island, Tasmania has been gaining a lot of tourists in the past 4 years. The increasing visitor economy resulted to increasing Airbnb listings as well. For first-time visitors, who wants to maximize their stay by going around the island, looking for multiple accommodations at different Tasmanian towns that satisfies their wants and needs in terms of amenities and accessibility to certain establishments is tedious and time-consuming. Using unsupervised clustering, this study aims to uncover similar Airbnb listings in Tasmania based on price, offerings, and amenities around the area.

A total of 5,087 active Airbnb listings in May 2020 were sourced and processed from InsideAirbnb and Open Street Maps using web-scraping tools. k-means algorithm was used to group the listings and internal validation metrics were utilized to determine the optimal number of clusters. It was determined that k = 3 is the optimal number of k using internal validation. It was found that price and room_type clearly sets the clusters apart. This algorithm can be used as mechanism for recommender systems where listings from the same clusters are suggested to tourists doing multiple bookings.