Traffic Congestion Analysis and Prediction Method based on XGBoost and LightGBM

In order to alleviate the problem of urban traffic congestion a traffic congestion analysis and prediction method based on XGBoost and LightGBM integration model is proposed in this paper. This method abandons the boosting-tree model widely used in traditional data mining, using a new weight-assigning algorithm is prompted, considering both the accuracy of data acquisition and the time complexity. The experimental results show that the mean square error (RMSE) of the prompted fused model remains about 2.5, and the predicted results are also consistent with the graphical results visualized by Echarts. The method has high prediction accuracy and plays a certain role in alleviating traffic pressure.