The operation of any intelligent transport systems requires efficient traffic forecasting, and hence, management of the system traffic and navigation of environmental terrains is improved. The availability of all the traffic environment information is of large extent. Spatio-temporal traffic data is dynamic and raises a lot of issues. This work is a model that combines the strengths of temporal convolutional networks and diffusion convolutional recurrent networks within a single architecture. TCN makes use of dilated convolution to add long time series, while DCRN performs a graph diffusion process allowing one to generate traffic distributions across the entire network at any instant of time. In contrast to such traditional approaches as LSTM and ARIMA, our hybrid SSTFN model is capable of producing results that are significantly better of all aspects of both short-term and long-term predictions made, based on the performance measures that were further compared with real-life data.