Analysis and visualization of accidents severity based on LightGBM-TPE

In recent years, road traffic accidents, as a leading cause of accidental deaths, have been attracting more and attention across several disciplines. Notably, the feature study on accidents severity can help exactly identify causality between different risk factors thereby substantially improving safety. Meanwhile, application data visualization to safety investigations is still lacking. Motivated by this, we incorporate method into machine learning analyze UK in 2017. A hybrid algorithm, namely Light Gradient Boosting Machine-Tree-structured Parzen Estimator (LightGBM-TPE) proposed. Compared with other typical algorithms, it performs better terms metrics f1,accuracy, recall precision. Using LightGBM-TPE calculate SHAP value each feature, find that “Longitude”, “Latitude”, “Hour” “Day_of_Week” are four most closely related accident severity. Visualization for further verifies this conclusion. Overall, our research tries explore an innovative way understand evaluate importance which suggest effective solutions improve • We propose model, understanding predicting accidents. utilize investigate Our model be adopted provide practical advices urban planners.