针对城市道路拥堵的优化随机森林预测模型

Optimized Random Forest Prediction Model for Urban Road Congestion

为了及时对城市道路拥堵情况进行预测,缓解通行压力、降低能源损耗,通过一种结合Spark与阴阳对优化随机森林的模型预测城市道路交通拥堵情况,利用阴阳对优化算法对随机森林进行参数调优,选取决策树个数和分裂属性个数最优解,以此建立阴阳对优化随机森林(Yin-Yang-pair optimization random forest,YYPORF)城市道路拥堵情况预测模型,并通过Spark实现了YYPORF模型的并行化设计方案。结果表明,基于Spark的YYPORF准确率达到95.58%,较传统随机森林提高了3.17%;加速比达到2.83。可见,所提出的模型预测性能更优,可为大数据背景下城市智能交通管理提供可靠依据。

In order to predict urban road congestion in time,alleviate traffic pressure and reduce energy consumption,a model combining Spark and Yin-Yang-pair optimization to optimize random forest was used to predict urban road traffic congestion.The Yin-Yang-pair optimization was used to optimize the parameters of random forest to select the optimal solution of the number of decision trees and the number of split attributes.Based on this,a Yin-Yang-pair Optimization Random Forest(YYPORF)urban road congestion prediction model is established,and the parallel design scheme of YYPORF model is implemented by Spark.The results show that the accuracy of YYPORF based on Spark is 95.58%,which is 3.17%higher than that of traditional random forest.And the speedup ratio reaches 2.83.Therefore,the model proposed has the best performance,and can provide a reliable basis for urban intelligent traffic management under the background of big data.