Analysis Model Integrating Traffic Volume and Temperature for Traffic Flow Prediction

Rising global vehicle numbers lead to traffic congestion, which is influenced by temperature. Nowadays, temperature forecasting has reached a high level of sophistication, so it is feasible to predict traffic volume based on the forecasting temperatures. To predict the traffic volume, collected the traffic and weather data of I-94 Interstate highway. Based on the data, the simple linear regression model (LR), polynomial regression model (PR) and random forest (RF) were applied. With the comparison between simulation performances, RF demonstrated the highest accuracy, with a correlation coefficient (R2) of 0.8364. To bolster the simulation performances of regression models, more constraints should be involved such as rainfall, snowfall, and cloud cover, should also be considered. Besides, more sophisticated algorithms, such as regression-enhanced random forests (RERFs) and improved random forest classifiers (RFC), should be applied as well. In conclusion, as more related factors are involved, the performance of regression models will be better, which serve for traffic management effectively, with accurate traffic flow prediction.