Traffic congestion during the peak commuting period is becoming more and more serious in many cities in China. Accurate prediction of traffic flow on urban roads during the peak commuting period is one of the key fundamental problems to reduce traffic congestion and build intelligent transportation. This paper proposes a low-cost and high-efficiency method to address the challenges of the existing ORIGIN-DESTINATION survey method, which is costly, inefficient and has limited accuracy of results. By using the city-wide traffic, social security, insurance, population and other digital city data, we group the residents' travel modes into four types: walking, public transportation, rail transit and self-driving based on the service radius and travel distance of transportation, and analyze and count the number of private car trips and rail transit passenger flow by using the rail transit priority principle and the shortest path algorithm, and predict the commuting peak period based on this method. The road traffic flow during the peak commuting period is also predicted. The validity and reliability of the method are demonstrated by using typical data of Wuhan city as an example.