Abstract: Based on the analysis result of the high no-linear and time-spatial coupling of the traffic flow between the detectors segment, data mining technology is used to extract the time-spatial data of traffic flow with the detector data and probe vehicle GPS data of the expressway in Shanghai.The K-deformed multinomial logit model is put forward to predict the traffic condition, and the characteristic parameters are used to setup the K-deformed multinomial logit model for the traffic condition prediction.The data validation of the expressway in Shanghai is simulated on the platform of VISSIM COM and Microsoft Visual C+ +6.0, and the results show that the precision of traffic condition prediction using the K-deformed multinomial logit model is 93. 46%, average travel time and vehicle delay reduce by 17.1% and 11.9% respectively, average vehicle speed improves by 14.6%.