摘要
探讨了部分线性回归模型当非参数分量受到单调条件限制时的估计问题,利用惩罚局部线性核估计方法给出参数分量与非参数分量αRLS、gRLS的估计,并进一步研究了^αRLS及^gRLS的渐进性质,结果表明:新得到的非参数约束条件下的估计是a.s.收敛的,并且保证了单调性和边界点适应性。将问题推广到一般情况,探讨了部分线性回归模型当非参数分量受到高阶导数约束时,参数分量的估计^αRLS。
Discusses the estimation problem of partially linear regression model when the nonparametric component is limited by monotonic constraints, by punishing local linear kernel estimation method gives parametric and nonparametrie component estimates： ams,g^s, and further researches the asymptotic properties of ams,gv, LS. The results show that： the new estimation of nonparametric constraints is convergence, and can ensure the monotonicity and boundary point adaptability. At last, the problem is extended to the general case; the estimation of the parameter components is constrained by the higher order derivative is discussed.
出处
《统计与信息论坛》
CSSCI
北大核心
2017年第7期30-35,共6页
Journal of Statistics and Information
基金
山东省社科规划基金项目<新常态下山东省人口老龄化对居民消费影响效应研究>(15DJJJ14)
山东省自然科学基金项目<基于随机内生增长理论框架的环境保护与可持续发展问题研究>(ZR2015PG006)
关键词
惩罚局部多项式
核方法
单调约束
半参数模型
渐进性质
local polynomial estimation
nuclear method
monotonous conditions
partially linearregression model
asymptotic properties
