摘要
传统泊松分布模型分析网页信息搜索错误概率过程中,模型构建完成后未检验模型优度,导致预测网页信息搜索错误概率的精度较低,以广义线性模型基本原理为前提,基于广义线性回归设计网页信息错误风险概率广义线性模型,采用极大似然方法估计模型回归参数α取值,通过SD方法、Pearsonχ~2方法检验模型拟合优度,分析模型变量弹性,获取影响网页信息搜索错误概率的关键因素,完成网页信息搜索错误概率分析.实验结果表明,所设计模型预测网页信息搜索错误概率精度均值高达98.5%,预测网页信息搜索错误次数与实际值吻合,能够得出影响网页信息搜索错误概率的因素.
The traditional Poisson distribution model does not test the goodness after the model is built, which leads to the low accuracy of the prediction probability. On the premise of the basic principle of generalized linear model and based on generalized linear regression, this paper designs the generalized linear model of web page information error risk probability, estimates the regression parameters of the model by maximum likelihood method, tests the goodness of fit by SD and Pearson χ^2 methods, analyses the elasticity of model variables, obtains the key factors affecting the search error probability, and completes the error probability analysis. The experimental results show that the average accuracy of the model is 98.5%. The number of prediction errors coincides with the actual value, and the factors affecting the error probability are obtained.
作者
黄秀常
HUANG Xiu-chang(Institute of Mechanical and Electrical Information,Yiwu Industrial and Commercial College, Yiwu Zhejiang 322000, China)
出处
《菏泽学院学报》
2019年第2期14-20,共7页
Journal of Heze University
基金
2019年度义乌工商职业技术学院科研项目(2019JD304-01)
关键词
广义线性模型
信息搜索
联系函数
极大似然估计
弹性系数
错误概率
generalized linear model
information search
connection function
maximum likelihood estimation
elastic coefficient
error probability
