Analysis and Prediction of Tokyo's Traffic Flow Using the SARIMA Time Series Forecasting Method

With the development of Intelligent Transportation Systems (ITS), accurate traffic flow forecasting has become crucial for urban traffic management and infrastructure planning. This study employs the Seasonal Autoregressive Integrated Moving Average (SARIMA) model for time series analysis and forecasting of traffic flow data within the Tokyo metropolitan area from 2015 to 2024. The data is sourced from the "Tokyo Traffic" dataset on the Kaggle platform, encompassing daily traffic records and associated variables for the C1 Loop Expressway. The analysis reveals a significant 12-month periodicity in Tokyo's traffic flow. The model parameters were identified as SARIMA (0,1,1) (0,1,0,12). Validation via the Augmented Dickey-Fuller (ADF) test and white noise tests confirmed that the residuals satisfy the model assumptions. The forecasting results demonstrate that the model achieves a Mean Absolute Percentage Error (MAPE) of 5.57%. This represents an approximate 45% improvement in accuracy compared to the traditional ARIMA model and outperforms benchmarks from the Seattle highway network (6.48%) and the California PEMS08 dataset (7.15%).This research provides a reliable data-driven forecasting analysis to support dynamic management strategies for Tokyo's traffic in the post-pandemic era, such as toll optimization and network dispatching.