Location based services APIs for measuring the attractiveness of long-term rental apartment location using machine learning model

Abstract

Smart cities produce high volumes of data that could be shared through APIs. This open data could be used to overcome the challenges of urbanization. This research aims at using the Location-based-services APIs to measure the attractiveness of a residence location for people who are searching for long-term rental apartments. The research uses the days on market as an indicator for residence attractiveness. This will occur through the developing a machine learning model to predict days on market whereas the urban features are part of the predictors. The research utilizes transaction data and profile information for long term rental apartments crawled from a real estate website. The research assumption has been tested using the principal component regression. Several machine learning models have been evaluated. The Gradient Boosting Regression is selected based on its performance. Shapley value is used to determine the comportment of the predictors in machine learning model. Although the urban feature component is a relatively weak predictor, but it is statically significant, whereas walkability, transit availability, and bikeability negatively influence the days on market while traffic congestion has a positive impact. These shows that renters prefer well serviced areas even when it is congested.