摘要
非洲国家的安全形势日益严峻,为了更好地服务中非合作,本文基于随机森林分类算法,构建一套安全风险预警模型,用于实现对冲突危险等级的预测。通过挖掘种族和矿产资源影响冲突的空间特征并将其数据化,使输入数据能反映冲突因素的作用机制,改进了有限数据下处理冲突预测问题的拟合效果。采用该模型对肯尼亚国内地理网格单位的冲突等级进行预测和结果评估,证实了指标体系模型的科学合理性。同时将其与决策树和BP神经网络进行效果对比,该模型的精确率、召回率以及F1指数均占优,证实该指标体系模型预测效果较好。
The security situation of African countries is becoming increasingly serious,in order to better serve China-Africa cooperation,this paper constructs a set of security risk early warning models based on the random forest classification algorithm,which is used to realize the prediction of the danger level of conflict.And race and mineral resources variables are mined to make the input data can reflect the conflict factor mechanism,improving the prediction performance.Through the prediction and result evaluation of the domestic geographic grid units in Kenya,the scientific rationality of the index system model is confirmed and its effects were compared with those of decision trees and BP neural networks,and the model's precision rate,recall rate and F1 index were superior,confirming that the model prediction of the index system is better.
作者
张永晴
Zhang Yongqing(School of International Relations,Beijing Foreign Studies University,Beijing 100081,China)
出处
《计算机时代》
2025年第1期36-41,共6页
Computer Era
关键词
中非合作
非洲国家
冲突预测
随机森林
GIS
预警模型
China-Africa cooperation
Africa
domestic conflict
Random Forest
GIS
Early warning model
