基于Logistic回归的犯罪概率预测研究

The Crime Probability Prediction Based on Logistic Regression

国内能够指导基层公安机关工作的犯罪微观预测研究甚少.基于Logistic回归模型构建了一个犯罪概率预测模型,对模型指标体系选取、数据预处理、模型训练等关键问题进行了重点研究,并进行了理论检验和实证分析证明.

There is little research on the crime prediction which can guide the work of grass-root public security bureau. Grounded on the logistic regression model,the paper attempts to construct a prediction model of crime probability by studying such key issues as the selection of index system of the model,the preprocessing of the data and the training in accordance with the model,etc. The model is theoretically tested and empirically verified.