Citation

Takahashi K, Miyazaki Y, Kitamura K, Sato F. Trans. Soc. Automot. Eng. Jpn. 2023; 55(1): 56-62.

Copyright

(Copyright © 2023, Society of Automotive Engineers of Japan)

DOI

10.11351/jsaeronbun.55.56

PMID

unavailable

Abstract

It is necessary to estimate the injury of occupants during car accidents to estimate the effect of injury reduction performance of autonomous driving systems. Although there are some estimation models of injury of occupants based on logistic regression, logistic regression has the problem of being unable to express nonlinear relationships between explanatory and objective variables. In this study, we used LightGBM, a decision tree model, and our own selected explanatory variables to construct an injury prediction model to predict the probability of VAIS3+ of vehicles. It showed a significant improvement in performance from URGENCY.

Language: ja

Keywords

Accident analysis; Gradient boosting decision tree; Injury prediction; Machine learning; Safety