The Demand Analysis and Forecast of APP-based Taxi Service via Machine Learning

The app-based taxi service has significantly changed the transportation market of the modern city in the last decade. It is very important to predict the demand of APP-based taxi service. In this paper, we proposed a study to analyze and forecast the demand of APP-based taxi service by using two different machine learning approaches, linear and sinusoidal regression. The Wujin District of Changzhou City in China is the study area in the experiment. The experimental result shows that the linear regression is better than the sinusoidal regression, although the request of APP-based taxi service have the periodic variation naturally. This can support the decision of the service platform and the cabdrivers.