In order to explore the phenomenon of informal intra-platform ties, among on-demand ride-hailing drivers in North America, Amaris, Ryan and Ziling analyzed the wordings of individual posts scraped from www.uberpeople.net, a widely used online platform for uber users, to detect whether a post gets emotional. They classified each post to be positive, negative or neutral by training three multi-class classifiers : Logistic Regression, Support Vector Machine and Random Forest, on binary bags of words ( the data representation of each post ) and obtained an accuracy of 73.97%, 71.17% and 63.23% for each model respectively.
Uberpeople
Sentiment analysis
Updates
Leave feedback in the comments!
Log in or sign up for Devpost to join the conversation.