In order to better solve the problem of unbalanced supply and demand of connected shared bikes, this paper takes shared bikes as the research object, analyzes the usage characteristics of connected bikes in different types of public transport stations, and puts forward a data-based feature extraction method of shared bikes. Firstly, the usage data of shared bikes were collected, and the starting and finishing points were decoded. The public transport stations were divided into five typical types according to the decoded longitude, latitude and surrounding land types. Secondly, the connectivity activity, connectivity distance and user loyalty are put forward as the characteristic indicators of bike-sharing travel. Finally, taking the bicycle data of Chaoyang District of Beijing as an example, the travel characteristic indexes of shared bikes are analyzed. The results show that, as the “last kilometer” travel connecting tool of public transport, the peak of the use of shared bikes connecting residential stations is 6:30 to 9:30, and that of other stations is 7:30 to 9:30. The connecting distance of shared bikes is generally less than 2km, but the connecting distance of office sites can reach 3km, and this site has the highest user loyalty.