共查询到18条相似文献,搜索用时 78 毫秒
1.
本文研究了生长曲线模型的Potthoff-Roy变换,给出了这种变换在矩阵范数最小规则下的最佳选择.进一步本文给出了利用协变量来改进Potthoff-Roy变换的途径并研究了协变量的选择及对参数估计的影响. 相似文献
2.
3.
在回归分析中,当因变量存在双侧截断时,已有的统计方法会使得回归模型的系数估计与变量选择产生偏差.本文提出一种适用于双侧截断回归模型的系数估计与变量选择方法,且该方法允许回归模型中自变量的个数随着样本量增大并趋于无穷而趋于无穷.该方法的主要思想是,提出一种Mann-Whitney型的损失函数来进行纠偏,随后加入自适应最小绝对收缩和选择算子(least absolute shrinkage and selection operator, LASSO)惩罚项来进行变量选择.本文同时设计一种迭代算法来实现损失函数的优化;且证明了所提出估计量的相合性与渐近正态性,还给出所提出变量选择方法的神谕性(oracle property).本文通过随机模拟展示所提出方法在有限样本量下的表现,并使用所提出方法分析一个天文学领域的实际数据集. 相似文献
4.
5.
本文研究测量误差模型的自适应LASSO(least absolute shrinkage and selection operator)变量选择和系数估计问题.首先分别给出协变量有测量误差时的线性模型和部分线性模型自适应LASSO参数估计量,在一些正则条件下研究估计量的渐近性质,并且证明选择合适的调整参数,自适应LASSO参数估计量具有oracle性质.其次讨论估计的实现算法及惩罚参数和光滑参数的选择问题.最后通过模拟和一个实际数据分析研究了自适应LASSO变量选择方法的表现,结果表明,变量选择和参数估计效果良好. 相似文献
6.
来源于不同总体的数据异质性较大,数据“零取值”较多且离散度大,可利用零膨胀泊松(ZIP)混合回归模型建模分析,然而混合模型中自变量较多.为了筛选出重要变量,本文利用自适应LASSO对ZIP混合回归模型进行变量选择,即在似然函数中加入惩罚项,再利用EM算法估计参数.通过模拟,验证了该方法在变量选择和参数估计中的有效性.同时,将ZIP混合回归模型应用于预测借贷失败次数的实际数据分析,筛选出对借贷失败有重要影响的因素.最后,通过比较各模型的预测效果,得到ZIP混合回归模型优于泊松(Poisson),负二项(NB)和ZIP回归模型. 相似文献
7.
本文主要研究分组数据分位数回归模型的变量选择和估计问题.为了充分反映数据的分组信息,需要假定每组数据的回归系数可以分解成共性部分和分组后的个性部分.为了进行变量筛选,本文提出分解系数的Lasso估计,并进一步提出了自适应Lasso估计.在处理相应优化问题时,采用了变换观测矩阵的方法简化问题求解.本文给出了自适应Lasso估计的Oracle性质证明,并且通过数值模拟研究展示了所提方法的有限样本表现.最后,将此方法应用到乳腺浸润癌致病基因的变量筛选上来展示所提方法的实际应用表现. 相似文献
8.
与传统的的媒体营销模式相比,搜索引擎广告因其精准和投入低等特点获得巨大成功。但已有的搜索引擎广告点击率模型不能有效解决数据量大及特征维度高的问题,使预测结果的准确性大打折扣。本文构建了一种基于LASSO变量选择方法的广告点击率预测模型,能有效克服现有广告点击率模型在处理数据高维性和稀疏性方面的不足。利用某公司的竞价数据对模型进行验证,结果表明影响广告点击率的关键因素是广告关键词中的商标信息、地域信息和每点击成本。该研究结果为企业制定搜索引擎广告营销策略提供一定的理论依据。 相似文献
10.
11.
Kenichi Satoh Mika Kobayashi Yasunori Fujikoshi 《Journal of multivariate analysis》1997,60(2):277-292
In this paper we consider the problem of selecting the covariables within individuals in the growth curve model. We propose two modifications ofAICandMIC(Cp-static), which have improvements on the bias properties. Asymptotic distributions of variable slection criteria are derived under a general situation where a polynomial growth curve of degreej0is approximately suitable. A simulation study is also given to gain some understanding on the small sample properties of these variable selection criteria 相似文献
12.
生长曲线分析是了解事物随时间的变化特点的分析热点.然而,传统的曲线拟合方法不适用通过重复测量得到的结构资料.采用轮廓设计矩阵运算虽然解决结构资料的生长曲线分析,但是相应的SAS程序冗长且难懂.将多水平模型应用于生长曲线的分析,通过实例分析介绍应用过程,同时给出简单明了的SAS程序. 相似文献
13.
14.
15.
In this article, we propose a new Bayesian variable selection (BVS) approach via the graphical model and the Ising model, which we refer to as the “Bayesian Ising graphical model” (BIGM). The BIGM is developed by showing that the BVS problem based on the linear regression model can be considered as a complete graph and described by an Ising model with random interactions. There are several advantages of our BIGM: it is easy to (i) employ the single-site updating and cluster updating algorithm, both of which are suitable for problems with small sample sizes and a larger number of variables, (ii) extend this approach to nonparametric regression models, and (iii) incorporate graphical prior information. In our BIGM, the interactions are determined by the linear model coefficients, so we systematically study the performance of different scale normal mixture priors for the model coefficients by adopting the global-local shrinkage strategy. Our results indicate that the best prior for the model coefficients in terms of variable selection should place substantial weight on small, nonzero shrinkage. The methods are illustrated with simulated and real data. Supplementary materials for this article are available online. 相似文献
16.
In the research it is frequently assumed
that the growth curve is a polynomial in time. In practice,
researchers mainly use higher-order polynomials to obtain more
precise estimates. But this method has many defects, such as the
model can be easily affected by outliers and the polynomial
hypothesis may be much strong in practice. So in this paper we first
proposed nonparametric approach, local polynomial, instead of
parametric method for estimation in growth curve model. We give the
nonparametric growth curve model, and its nonparametric estimation.
Then discuss the large sample character of local polynomial
estimate. The ideal theoretical choice of a local bandwidth is also
discussed in detail in this paper. Finally, through the simulation
study, from the fitting curve and average square error box plot we
can clearly see that the performance of nonparametric approach is
much better than parametric technique. 相似文献
17.
增长曲线模型中两个最优线性预测 总被引:3,自引:0,他引:3
对一般增长曲线模型中一类线性可预测变量KY0L和φ-可预测变量的最优预测进行了研究,在一定条件下分别得到了最优线性无偏预测和最优φ-线性无偏预测,并证明了它们在几乎处处意义下的唯一性. 相似文献