Regression Model for Bike-Sharing Service by Using Machine Learning

Abstract

The bike sharing system has brought wide convenience to residents in the city and serves as important tools to transport from one place to another place. For the bike sharing companies, they need to know the total users of bike, so they can release suitable number of bikes into the market. This paper uses visualization technology to visualize data and figure out the possible factors which can impact the total number of users. After completing the data analyzing, this paper figures out the season, weather sit, feeling temperature, humanity and wind speed are the main factors which can have impacts on the total number of users. In the second stages, this paper uses regression model, NN model, ELM model and DELM model to predict the possible number of bike users. The input factors are season, weather sit, feeling temperature, humanity and wind speed. By analyzing regression model results, the ELM model has the best prediction, which can be used for real practice.