This paper studies the setting method of urban bus lane, and takes Guiyang City as the experimental city, which integrates the data of electric police, checkpoint passing, bus GPS, bus line, bus station distribution, road basic information, electric police and checkpoint geographic information. Build the network topology of video detection equipment, analyze the relationship between bus and social vehicle speed, build a feature selection model, and determine which sections of the city need to set up bus lanes. Then the model optimization algorithm is designed to further determine whether each road section is suitable for setting bus lanes and how to set them. Finally, a set of methods that can automatically calculate the setting scheme of urban bus lanes based on data is developed.