论文部分内容阅读
提出一种适用于多影响因素回归拟合的全社会客运量预测的点积-平移型支持向量机算法.该算法能够全面、系统地分析影响客运量需求变化的关联因素,通过关联元素去冗处理,确定作为支持向量机算法输入变量的核心关联元素.考虑到客运量预测是一个基于时序变化的外推过程,并且受社会经济发展中多项影响因素的制约,数据变化存在着阶段性特征,提出利用点积-平移型核函数来拟合需求变化过程.对历史数据集的测试结果表明,该算法性能评价满足要求,可为远景客运量预测提供理论依据.
This paper proposes a dot-product-translation SVM algorithm which is applicable to the whole society passenger traffic forecasting fitted by the regression model of multiple influencing factors.It can comprehensively and systematically analyze the related factors that affect the passenger demand changes, Processing, to determine the kernel-related elements of input variables as support vector machine algorithm.Considering that passenger volume forecasting is a process of extrapolation based on time-series changes and is subject to many factors in social and economic development, there are phased features , Proposed the use of dot-product-shift kernel function to fit the process of demand change.The test results of historical data sets show that the performance evaluation of the algorithm meets the requirements and can provide a theoretical basis for long-range passenger traffic forecasting.