Hit-and-Run Accidents Missing Data Interpolation Using Random Forest

Investigating the characteristics and regularities of historical accident data is a crucial pathway in researching road traffic accidents and traffic safety. However, the prevalence of missing data is an inescapable flaw in research on road traffic accidents. Multiple imputation theories and methodologies offer a potential solution for utilizing samples with missing data more effectively. This paper focuses on escapement accidents as the subject of study and establishes the FCS (Flexible Combination Strategy) multiple imputation strategy and process based on a Random Forest method model. It determines the number of multiple imputations for variables with different types and scales of missingness. On this basis, the paper compares the conclusions of the analysis of the full sample without missing data to derive the characteristics of factors influencing the occurrence of escapement accidents under the multiple imputation model. It also analyzes and discusses the similarities and differences of the analytical conclusions. This research provides guidance and reference for improving the road traffic accident data system, preventing and reducing the occurrence of road traffic accidents, and reducing the severity of injuries caused by accidents.