In this paper，we propose an approach for forecasting traffic flow，which employs first a double exponential smoothing(DES)model to predict thefuture datum．Then，the DES model is used again for making a residual correction scheine．Furthermore，a Markov forecast model is adopted to construct aforecast trend adj ustment scheme．Smoothing parameters of the DES model is determined by using the genetic algorithm and the gradient descent algorithm．TheDrediction model is applied to forecasting the traffic flow of Shiji Route in ZiboCitv，the forecast resuIts of proposed model are compared with that of ARIMA．