基于大数据的服刑人员危险性预测

在对监狱服刑人员再犯罪预测与危险性评估应用现状进行分析的基础上,提出了一种基于大数据的监狱服刑人员危险性识别与预测架构体系.在该体系的模型层,针对不同的价值密度、不同的数据类型,重点对架构中的统计模型、离群点检测模型、集成分类模型在服刑人员危险性识别与预测中的应用算法进行了描述,尤其是使用R软件包实验了服刑人员危险性集成分类识别与预测,并给出了分类预测误差.基于大数据的服刑人员危险性识别与预测体系可实现对服刑人员危险性的个性化、精准化预警,为大数据时代监狱的监管安全提供了可靠保障.

Prediction of the risk of offenders based on big data

Based on the analysis of current status of offenders recidivism prediction and risk assessment,we proposed an architecture system of identification and prediction of the risk of offenders based on big data.Aiming at different value density and different data types,in the model layer of the system,we described application algorithm of identification and prediction of offenders risk for the statistical model,outlier detection model,integrated classification model of architecture system. In particular,using R software package,we conducted integrated classification identification and prediction experiments,and gives the classification prediction error.The architecture system of identification and prediction of the risk of offenders is based on big data,can realize personalized and accurate early warning for offenders risk,and provide a reliable guarantee for the safety of offenders supervision in the big data era.