Urban Traffic Accident Factor Analysis, Risk Prediction and Early Warning Based on Big Data Technology

Urban road traffic safety has become a social problem related to people's life safety, the main factors affecting the occurrence and severity of road traffic accidents are human factors, time, month, vehicle conditions, lighting, road conditions, weather conditions, visibility, etc. Using descriptive and predictive modeling, descriptive modeling is based on Hive database software to analyze the full sample data or Pyechart to visualize the data, the model studies and charts on historical data to obtain corresponding data regularity. Predictive models use machine learning algorithms, accident severity is taken as the target variable, through data cleaning, PCA was used for dimension reduction, RobustScaler data normalization and other processing, select the feature variables, Random forest algorithm or Xgboost algorithm was used to analyze the importance of characteristic variables that affect the severity of accidents. So as to find out the rules, in order to take preventive measures, reduce traffic accidents, to ensure people's safety.