A Comparison of Logistic Regression, Classification and Regression Tree, and Neural Networks Models in Predicting Violent Re-Offending

被引:70
作者
Liu, Yuan Y. [1 ]
Yang, Min [1 ,2 ]
Ramsay, Malcolm [3 ]
Li, Xiao S. [1 ]
Coid, Jeremy W. [4 ]
机构
[1] Sichuan Univ, Sch Publ Hlth, Dept Hlth Stat, Chengdu 610041, Peoples R China
[2] Univ Nottingham, Sch Community Hlth Sci, Div Psychiat, Nottingham NG7 2TU, England
[3] Minist Justice, Partnerships & Hlth Strategy Unit, London SW1H 9AJ, England
[4] Queen Mary Univ London, Forens Psychiat Res Unit, London EC1A 7BE, England
关键词
Violence reconviction; Risk assessment; Neural networks; Classification and regression tree; HCR-20; RISK-ASSESSMENT; PSYCHIATRIC-PATIENTS; STATISTICAL-MODELS; PCL-R; RECIDIVISM; ACCURACY; HCR-20; HEALTH; STRATIFICATION; CONSEQUENCES;
D O I
10.1007/s10940-011-9137-7
中图分类号
DF [法律]; D9 [法律];
学科分类号
0301 ;
摘要
Previous studies that have compared logistic regression (LR), classification and regression tree (CART), and neural networks (NNs) models for their predictive validity have shown inconsistent results in demonstrating superiority of any one model. The three models were tested in a prospective sample of 1225 UK male prisoners followed up for a mean of 3.31 years after release. Items in a widely-used risk assessment instrument (the Historical, Clinical, Risk Management-20, or HCR-20) were used as predictors and violent reconvictions as outcome. Multi-validation procedure was used to reduce sampling error in reporting the predictive accuracy. The low base rate was controlled by using different measures in the three models to minimize prediction error and achieve a more balanced classification. Overall accuracy of the three models varied between 0.59 and 0.67, with an overall AUC range of 0.65-0.72. Although the performance of NNs was slightly better than that of LR and CART models, it did not demonstrate a significant improvement.
引用
收藏
页码:547 / 573
页数:27
相关论文
共 102 条
[71]   Applying artificial neural network models to clinical decision making [J].
Price, RK ;
Spitznagel, EL ;
Downey, TJ ;
Meyer, DJ ;
Risk, NK ;
El-Ghazzawy, OG .
PSYCHOLOGICAL ASSESSMENT, 2000, 12 (01) :40-51
[72]  
Quinlan J. R., 1993, C4.5: Programs for Machine Learning
[73]   Improved use of continuous attributes in C4.5 [J].
Quinlan, JR .
JOURNAL OF ARTIFICIAL INTELLIGENCE RESEARCH, 1996, 4 :77-90
[74]   THE ACCURACY, EFFICIENCY, AND VALIDITY OF A PREDICTION INSTRUMENT [J].
Reiss, Albert J., Jr. .
AMERICAN JOURNAL OF SOCIOLOGY, 1951, 56 (06) :552-561
[75]   Comparing effect sizes in follow-up studies:: ROC area, Cohen's d, and r [J].
Rice, ME ;
Harris, GT .
LAW AND HUMAN BEHAVIOR, 2005, 29 (05) :615-620
[76]   VIOLENT RECIDIVISM - ASSESSING PREDICTIVE-VALIDITY [J].
RICE, ME ;
HARRIS, GT .
JOURNAL OF CONSULTING AND CLINICAL PSYCHOLOGY, 1995, 63 (05) :737-748
[77]  
Ripley B.D., 1996, Pattern Recognition and Neural Network, DOI [10.1017/CBO9780511812651, DOI 10.1017/CBO9780511812651]
[78]   Assessing violence risk in stalking cases: A regression tree approach [J].
Rosenfeld, B ;
Lewis, C .
LAW AND HUMAN BEHAVIOR, 2005, 29 (03) :343-357
[79]   Factors associated with violence in stalking and obsessional harassment cases [J].
Rosenfeld, B ;
Harmon, R .
CRIMINAL JUSTICE AND BEHAVIOR, 2002, 29 (06) :671-691
[80]  
Rumelhart D.E., 1985, Parallel Distributed Processing: Explorations in the Microstructure of Cognition, DOI [10.1016/b978-1-4832-1446-7.50035-2, DOI 10.7551/MITPRESS/5236.001.0001]