Monte Carlo EM梯度法在MATLAB中的实现

文章使用空间广义线性混合模型为连续空间非正态变量建模,在MATLAB中实现模型参数估计的MCEMG算法,即结合Monte Carlo样本的EM梯度法,求解参数的极大似然估计及采样点随机效应的最小均方误估计。在GS+中进行随机效应的普通克里格插值,并最终对非采样点响应变量进行预测。模拟仿真结果显示该方法参数估计与真实值较接近,响应变量预测结果能反应真实数据总体分布情况。     

固定效应部分线性变系数面板模型的快速有效估计

本文将最小二乘支持向量机(LSSVM) 和二次推断函数法(QIF) 相结合，为个体内具有相关结构的固定效应部分线性变系数面板模型提供了一种新的快速估计方法；在一定的正则条件下，论证了参数估计量的渐近正态性和非参数估计量的收敛速度；采用Monte Carlo模拟考察了估计方法在有限样本下的表现并将估计技术应用于现实数据分析。该方法不仅保证了估计的有效性和统计推断力，而且程序运行速度得到较大幅度提升。     

基于多因子选股的半监督核聚类算法改进研究

研究了一种附有引力影响因子的半监督K-means核函数聚类算法,并将该方法应用于多因子选股模型中。研究表明,相比传统的聚类模型,改进的模型具有较强的泛化能力,模型在处理样本线性不可分、样本分布非球状簇等问题上具有明显的优势,能选出较优的股票组合。     

工程质量政府监督多层次利益分配与激励协同机制探究

The implementation of the government supervision of the quality of the project is an international practice. The basic form of government supervision of engineering quality is government supervision on the quality behavior of the engineering main bodies and its results by the competent government department entrusted. Its essence is a dual principal-agent process. The frequent accidents of the engineering quality reflect the loss and failure of the government law enforcement supervision of the engineering quality to some extent. Its root lies in the lack of endogenous power in the law enforcement supervision of the project quality government supervisors in the law enforcement supervision. Therefore, the incentive coordination mechanism of the government supervision based on the multi-level interest distribution is worth explored. In views of the multi-level management system which is formed by the government departments, government quality supervision organizations, quality supervision team (or group) for the government supervision of engineering quality. The benefit distribution function between every party is constructed, and the game model of the multi-level incentive and coordination for the government supervision in engineering quality is built. To solve and deduce from the first stage of the cooperative game and the second stage of the non-cooperative game, the cooperative game can obtain the reward coefficient: . The coordination degree of the best effort can be obtained by the non cooperative game. The result shows that:the coordination degree of government engineering quality supervisor is related to the coordination costs, and had nothing to do with fixed costs. The benefit distribution coefficient not only depends on the efforts of the quality government monitors, but also on the efficiency of other parties' efforts. The quality supervisors of the project will also focus on the coordination with other parties when enhancing their management capabilities to improve the overall performance of project quality government supervision. The strategy of the incentive coordination mechanism for the supervision and cooperation of the project quality government is:the government quality supervision team should set up the supervisory team properly, improve the coordination efficiency and reduce the cost of supervision-coordination to maximize the value of self-motivation. Quality supervision team (or group) should establish the partnership to improve the coordination efficiency for achieving the maximization of their own incentive value.The model and conclusion of incentive synergy mechanism based on multi-level benefit allocation mechanism are researched. It can provide theoretical support and practice reference for the market governance and supervision of general public goods.     

函数与方程思想方法在中考解题中的应用

函数与方程思想方法是中学数学的基本思想，几乎渗透到中学数学的各个领域，在解题中有着广泛的应用。函数思想与方程思想虽有着各自的概念与性质，但它们之间又密切相关。本文将对函数与方程思想的概念、如何相互转化以高效解题及其在中考中的应用加以详细阐述。     

向量小波和正交小波尺度函数的设计及证明

文章通过多分辨分析理论和仿酋矢量滤波器组理论,讨论正交小波的存在,提出了组建一系列正交小波数据包的算法,得到了利用小波包组建L2(R,Cμ)(2≤μ∈Z)空间的多种标准正交基的一些方法.     

中国货币冲销政策选择有效性的实证研究

中国国际收支双顺差,且长期偏高。在现行外汇管理体制下,外汇储备的增加导致外汇占款增加,基础货币被动投放。这不仅干扰了中国货币政策独立性,也加速了国内通货膨胀。本文运用2001年1月至2011年12月的中国宏观经济月度数据,对中国货币冲销有效性作了估计。研究发现,受国债市场不发达和汇率制度等因素限制,货币工具部分冲销了样本期内央行被动投放的基础货币。     

《数理统计》教学中的若干思考

文章在总结了《数理统计》多年教学的基础上,提出在教学中应注意的几点问题,主要包括:统计分布的标准化、似然函数的构造、最小方差无偏估计以及C-R不等式等。     

多维copula函数的一种随机变量模拟生成方法

在对Archimedean copula函数性质分析的基础上,文章给出了多维copula函数的分布函数的一种计算方法,此方法的给出为多维copula函数的模型选择提供了方便,另外,基于多维copula函数的分布函数的表达式又引出了一种三维随机变量模拟生成的方法,并且通过实证分析显示此方法给出的模拟值和真实值较为接近.