基于ARIMA-LSSVM混合模型的犯罪时间序列预测  

A hybrid ARIMA-LSSVM model for crime time series forecast

作　　者：涂小萌 陈强国[1] 

机构地区：[1]武汉大学计算机学院 国家多媒体软件工程研究中心,湖北武汉430072

出　　处：《电子技术应用》2015年第2期160-162,166,共4页Application of Electronic Technique

基　　金：湖北省重大科技创新计划项目(2013AAA020)

摘　　要：对犯罪时间序列的预测对帮助公安部门更好地掌握犯罪动态,实现智能犯罪发现具有重大意义。针对犯罪时间序列预测的计算需求,结合真实犯罪数据集,提出了ARIMA-LSSVM混合模型。该模型通过ARIMA预测出时间序列的线性部分,通过PSO优化的LSSVM模型预测非线性部分,以对序列进行充分拟合,最后通过混合算法计算最终结果。使用此混合模型达到了精准的预测效果,证明了模型的有效性。Forecasting the time series of crime can help public security department better grasp the criminal dynamic and has a great significance in realizing intelligent crime detection. Specific to the computing need of the prediction of crime time series, combining real crime dataset, a hybrid ARIMA-LSSVM model is put forward. This model utilizes ARIMA model predicting the linear component of series and PSO optimized LSSVM model predicting the nonlinear component to fully fitting the series, and computes the final result by using the hybrid algorithm. The great efficiency and accuracy are reached, which proves the validity of the model.

关 键 词：犯罪时间序列 相空间重构 滑动自回归平均模型 后向传播神经网络 PSO-LSSVM 

分 类 号：P391[天文地球—地球物理学] D917.9[政治法律—法学]