Brand- and price-based usage pattern analysis on peer-to-peer car-sharing platform

Abstract

The Sharing Economy has greatly improved people’s lives by bringing convenience of sharing daily necessities from portable chargers to vehicles such as bicycles and cars with lower cost. With the prevalence of Sharing Economy and the increasing support on car- sharing from the Chinese government, the NTU-BMW Future Mobility Research Lab wants to find out more on consumers’ usage patterns, based on car brand and price of the Chinese peer-to-peer car-sharing users. This project uses the data crawled from the START peer-to-peer car-sharing platform in China. Data cleaning, feature extraction and creation, K-fold, correlation analysis and random forest classification are implemented in this project to perform the analysis. Data cleaning includes data pre-processing and removing non-numerical characters and incomplete data. The processed data are used to produce demographics illustrated using line and bar graphs, showing different point of views based on the attributes used. The results consist of the demographics of the dataset used, categorized by different attributes such as car age, mileage and car brand. Analyses are made based on the correlation heatmaps produced and the important features listed by random forest classifier. This can help future researches to find out the underlying social or emotional factors consumers have to make a decision to make rent or a purchase.