의사결정나무 분석기법을 활용한 성범죄의 집행유예 결정인자 연구

A study on the impact of considerations factors on suspended sentence for sexual offence using decision tree analysis method

The suspended sentence has been most commonly used as a criminal penalty in Korean criminal justice system. Only small number of empirical research have examined the impact of consideration factors on suspended sentence. This study have examined sentences for crimes of sexual assault using data mining technique(decision tree analysis). 301 criminal cases were analyzed by logistic regression which has been applied to predicting legal decision making, with an alternative method-decision trees. The finding using logistic regression showed the consideration factors such victim s opposition to punishment, no criminal history, sincere remorse were significant to predict a possibility of suspended sentence for sexual crimes. Contrarily, decision tree analysis produced a simple and equally effect model that only two factors, victim’s opposition to punishment and sincere remorse, were significant to predict suspended sentence. Comparison between two method using AUC analysis showed no difference. Decision tree analysis is as good as logistic regression analysis in predicting suspended sentence and better for interpretation. Limitations and implications for policy and future research were discussed.