基于机器学习的故意伤害案件风险分析

Risk Analysis of Assaults Based on Machine Learning

当前,全国公安机关正在大力开展扫黑除恶行动,打击暴力犯罪是扫黑除恶的重要手段。对故意伤害案件这种典型的暴力案件进行风险分析能预测案件的风险后果和探测影响犯罪的风险因素,进而为公安机关决策提供帮助。基于机器学习方法,利用A市2014-2016年故意伤害案件真实数据,通过特征筛选、分类等特征工程方法,最后得到F1为0.72的随机深林模型。依据特征重要度排序发现,涉案人数、作案手段、案发地的周边环境是影响一起故意伤害案件的重要因素。因此在公安机关接收到与上述风险因素相关警情时应加大警力和装备,及时赶赴现场控制局面。

At present,public security organs across the country are vigorously launching anti-crime operations.Combating violent crimes is an impor⁃tant means.Performing a risk analysis on a typical violent case such as an intentional injury case can predict the risk consequences of the case and detect the risk factors that affect the crime,thereby providing assistance to the decision-making of public security organs.Based on the machine learning method,using the real data of intentional injury cases in City A from 2014 to 2016,through feature selection,clas⁃sification and other feature engineering methods,a random deep forest model with F1 of 0.72 is finally obtained.According to the ranking of the characteristics and importance,it was found that the number of people involved,the means of committing the crime,and the surround⁃ing environment of the place where the crime occurred were important factors affecting a case of intentional injury.Therefore,when the pub⁃lic security organ receives the police situation related to the above-mentioned risk factors,it should increase the police force and equip⁃ment,and rush to the scene to control the situation in a timely manner.