Data-driven means the control and decision making is based on the state of the data. The data-drive urban traffic flow guidance system is a decision making system which is based on large amount of traffic flow data. This paper gives detailed discussion of data-drive urban traffic flow guidance system and performs qualitative analysis on the main components of this system, which includes the distribution of traffic flow data collectors, the layers of guidance system, the feature vector of traffic flow state, and the comparison of traffic flow estimation techniques. The paper represents a new approach to study and construct the traffic guidance system.