### 统计代写|似然估计作业代写Probability and Estimation代考|Introduction to Ridge Regression

statistics-lab™ 为您的留学生涯保驾护航 在代写似然估计Probability and Estimation方面已经树立了自己的口碑, 保证靠谱, 高质且原创的统计Statistics代写服务。我们的专家在代写似然估计Probability and Estimation代写方面经验极为丰富，各种代写似然估计Probability and Estimation相关的作业也就用不着说。

• Statistical Inference 统计推断
• Statistical Computing 统计计算
• (Generalized) Linear Models 广义线性模型
• Statistical Machine Learning 统计机器学习
• Longitudinal Data Analysis 纵向数据分析
• Foundations of Data Science 数据科学基础

## 统计代写|似然估计作业代写Probability and Estimation代考| Introduction to Ridge Regression

Consider the common multiple linear regression model with the vector of coefficients, $\boldsymbol{\beta}=\left(\beta_{1}, \ldots, \beta_{p}\right)^{\top}$ given by
$$Y=X \boldsymbol{\beta}+\boldsymbol{\epsilon},$$
where $Y=\left(y_{1}, \ldots, y_{n}\right)^{\top}$ is a vector of $n$ responses, $X=\left(\boldsymbol{x}{1}, \ldots, \boldsymbol{x}{n}\right)^{\top}$ is an $n \times p$ design matrix of rank $p(\leq n), x_{i} \in \mathbb{R}^{p}$ is the vector of covariates, and $\boldsymbol{e}$ is an $n$-vector of independently and identically distributed (i.i.d.) random variables (rv.).

The least squares estimator (LSE) of $\boldsymbol{\beta}$, denoted by $\tilde{\boldsymbol{\beta}}{n}$, can be obtained by minimizing the residual sum of squares (RSS), the convex optimization problem, $$\min {\beta}\left{(\boldsymbol{Y}-\boldsymbol{X} \boldsymbol{\beta})^{\top}(\boldsymbol{Y}-\boldsymbol{X} \boldsymbol{\beta})\right}=\min _{\boldsymbol{\beta}}{S(\boldsymbol{\beta})},$$
where $S(\boldsymbol{\beta})=\boldsymbol{Y}^{\top} \boldsymbol{Y}-2 \boldsymbol{\beta}^{\top} \boldsymbol{X}^{\top} \boldsymbol{Y}+\boldsymbol{\beta}^{\top} \boldsymbol{X}^{\top} \boldsymbol{X} \boldsymbol{\beta}$ is the RSS. Solving
$$\frac{\partial S(\boldsymbol{\beta})}{\partial \boldsymbol{\beta}}=-2 \boldsymbol{X}^{\top} \boldsymbol{Y}+2 \boldsymbol{X}^{\top} \boldsymbol{X} \boldsymbol{\beta}=\mathbf{0}$$

with respect to (w.r.t.) $\beta$ gives
$$\tilde{\boldsymbol{\beta}}{n}=\left(\boldsymbol{X}^{\top} \boldsymbol{X}\right)^{-1} \boldsymbol{X}^{\top} \boldsymbol{Y} \text {. }$$ Suppose that $\mathbb{E}(\boldsymbol{\varepsilon})=0$ and $\mathbb{E}\left(\boldsymbol{\varepsilon} \boldsymbol{e}^{\top}\right)=\sigma^{2} \boldsymbol{I}{n}$ for some $\sigma^{2} \in \mathbb{R}^{+}$. Then, the variance-covariance matrix of LSE is given by
$$\operatorname{Var}\left(\tilde{\boldsymbol{\beta}}_{n}\right)=\sigma^{2}\left(\boldsymbol{X}^{\top} \boldsymbol{X}\right)^{-1}$$
Now, we consider the canonical form of the multiple linear regression model to illustrate how large eigenvalues of the design matrix $\boldsymbol{X}^{\top} \boldsymbol{X}$ may affect the efficiency of estimation.

Write the spectral decomposition of the positive definite design matrix $\boldsymbol{X}^{\top} \boldsymbol{X}$ to get $\boldsymbol{X}^{\top} \boldsymbol{X}=\boldsymbol{\Gamma} \boldsymbol{\Lambda} \boldsymbol{\Gamma}^{\top}$, where $\boldsymbol{\Gamma}(p \times p)$ is a column orthogonal matrix of eigenvectors and $\boldsymbol{\Lambda}=\operatorname{Diag}\left(\lambda_{1}, \ldots, \lambda_{p}\right)$, where $\lambda_{j}>0, j=1, \ldots, p$ is the ordered eigenvalue matrix corresponding to $\boldsymbol{X}^{\top} \boldsymbol{X}$. Then,
$$Y=T \xi+\epsilon, \quad T=X \Gamma, \quad \xi=\Gamma^{\top} \beta$$
The LSE of $\xi$ has the form,
\begin{aligned} \tilde{\xi}{n} &=\left(\boldsymbol{T}^{\top} \boldsymbol{T}^{-1} \boldsymbol{T}^{\top} \boldsymbol{Y}\right.\ &=\boldsymbol{\Lambda}^{-1} \boldsymbol{T}^{\top} \boldsymbol{Y} \end{aligned} The variance-covariance matrix of $\tilde{\xi}{n}$ is given by
$$\operatorname{Var}\left(\tilde{\xi}{n}\right)=\sigma^{2} \boldsymbol{\Lambda}^{-1}=\sigma^{2}\left[\begin{array}{cccc} \frac{1}{\lambda{1}} & 0 & \cdots & 0 \ 0 & \frac{1}{\lambda_{2}} & 0 & 0 \ \vdots & \cdots & \ddots & \vdots \ 0 & 0 & \cdots & \frac{1}{\lambda_{p}} \end{array}\right]$$
Summation of the diagonal elements of the variance-covariance matrix of $\tilde{\xi}{n}$ is equal to $\operatorname{tr}\left(\operatorname{Var}\left(\tilde{\xi}{n}\right)\right)=\sigma^{2} \sum_{j=1}^{p} \lambda_{j}^{-1}$. Apparently, small eigenvalues inflate the total variance of estimate or energy of $X^{\top} X$. Specifically, since the eigenvalues are ordered, if the first eigenvalue is small, it causes the variance to explode. If this happens, what must one do? In the following section, we consider this problem. Therefore, it is of interest to realize when the eigenvalues become small.

Before discussing this problem, a very primitive understanding is that if we enlarge the eigenvalues from $\lambda_{j}$ to $\lambda_{j}+k$, for some positive value, say, $k$, then we can prevent the total variance from exploding. Of course, the amount of recovery depends on the correct choice of the parameter, $k$.

## 统计代写|似然估计作业代写Probability and Estimation代考| Multicollinearity Problem

Multicollinearity or collinearity is the existence of near-linear relationships among the regressors, predictors, or input/exogenous variables. There are terms such as exact, complete and severe, or supercollinearity and moderate collinearity. Supercollinearity indicates that two (or multiple) covariates are linearly dependent, and moderate occurs when covariates are moderately correlated. In the complete collinearity case, the design matrix is not invertible. This case mostly occurs in a high-dimensional situation (e.g. microarray measure) in which the number of covariates $(p)$ exceeds the number of samples $(n)$.

Moderation occurs when the relationship between two variables depends on a third variable, namely, the moderator. This case mostly happens in structural equation modeling. Although moderate multicollinearity does not cause the mathematical problems of complete multicollinearity, it does affect the

interpretation of model parameter estimates. According to Montgomery et al. (2012), if there is no linear relationship between the regressors, they are said to be orthogonal.

Multicollinearity or ill-conditioning can create inaccurate estimates of the regression coefficients, inflate the standard errors of the regression coefficients, deflate the partial $t$-tests for the regression coefficients, give false and nonsignificant $p$-values, and degrade the predictability of the model. It also causes changes in the direction of signs of the coefficient estimates. According to Montgomery et al. (2012), there are five sources for multicollinearity: (i) data collection, (ii) physical constraints, (iii) overdefined model, (iv) model choice or specification, and (v) outliers.

There are many studies that well explain the problem of multicollinearity. Since theoretical aspects of ridge regression and related issues are our goal, we refer the reader to Montgomery et al. (2012) for illustrative examples and comprehensive study on the multicollinearity and diagnostic measures such as correlation matrix, eigen system analysis of $\boldsymbol{X}^{\top} \boldsymbol{X}$, known as condition number, or variance decomposition proportion and variance inflation factor (VIF). To end this section, we consider a frequently used example in a ridge regression, namely, the Portland cement data introduced by Woods et al. (1932) from Najarian et al. (2013). This data set has been analyzed by many authors, e.g. Kaciranlar et al. (1999), Kibria (2003), and Arashi et al. (2015). We assemble the data as follows.

## 统计代写|似然估计作业代写Probability and Estimation代考| Ridge Regression Estimator: Ridge Notion

If the regression coefficients $\beta_{j} s$ are unconstrained, then they can explode (become large); this results in high variance. Hence, in order to control the variance, one may regularize the regression coefficients and determine how large the coefficient grows. In other words, one may impose a constraint on them so as not to get unboundedly large or penalized large regression coefficients. One type of constraint is the ridge constraint given by $\sum_{j=1}^{p} \beta_{j}^{2} \leq t$ for some positive value $t$. Hence, the minimization of the penalized residual sum of squares (PRSS) is equivalent to solving the following convex optimization problem,
$$\min {\beta}\left{(Y-X \boldsymbol{\beta})^{\top}(\boldsymbol{Y}-\boldsymbol{X} \boldsymbol{\beta})\right} \quad \text { such that } \sum{j=1}^{p} \beta_{j}^{2} \leq t$$
for some positive value $t$.
In general, the PRSS is defined by
$$(\boldsymbol{Y}-\boldsymbol{X} \boldsymbol{\beta})^{\top}(\boldsymbol{Y}-\boldsymbol{X} \boldsymbol{\beta})+k|\boldsymbol{\beta}|^{2}, \quad|\boldsymbol{\beta}|^{2}=\sum_{j=1}^{p} \beta_{j}^{2}$$
Since the PRSS is a convex function w.r.t. $\boldsymbol{\beta}$, it has a unique solution. Because of the ridge constraint, the solution is termed as the ridge regression estimator (RRE).

To derive the RRE, we solve the following convex optimization problem
$$\min {\boldsymbol{\beta}}\left{(\boldsymbol{Y}-\boldsymbol{X} \boldsymbol{\beta})^{\top}(\boldsymbol{Y}-\boldsymbol{X} \boldsymbol{\beta})+k|\boldsymbol{\beta}|^{2}\right}=\min {\boldsymbol{\beta}}{\operatorname{PS}(\boldsymbol{\beta})},$$
where $\operatorname{PS}(\boldsymbol{\beta})=\boldsymbol{Y}^{\top} \boldsymbol{Y}-2 \boldsymbol{\beta}^{\top} \boldsymbol{X}^{\top} \boldsymbol{Y}+\boldsymbol{\beta}^{\top} \boldsymbol{X}^{\top} \boldsymbol{X} \boldsymbol{\beta}+k \boldsymbol{\beta}^{\top} \boldsymbol{\beta}$ is the PRSS. Solving
$$\frac{\partial \mathrm{PS}(\boldsymbol{\beta})}{\partial \boldsymbol{\beta}}=-2 \boldsymbol{X}^{\top} \boldsymbol{Y}+2 \boldsymbol{X}^{\top} \boldsymbol{X} \boldsymbol{\beta}+2 k \boldsymbol{\beta}=\mathbf{0}$$
w.r.t. $\beta$ gives the RRE,
$$\hat{\boldsymbol{\beta}}{n}^{\mathrm{RR}}(k)=\left(\boldsymbol{X}^{\top} \boldsymbol{X}+k \boldsymbol{I}{p}\right)^{-1} \boldsymbol{X}^{\top} \boldsymbol{Y} .$$
Here, $k$ is the shrinkage (tuning) parameter. Indeed, $k$ tunes (controls) the size of the coefficients, and hence regularizes them. As $k \rightarrow 0$, the RRE simplifies to the LSE. Also, as $k \rightarrow \infty$, the RREs approach zero. Hence, the optimal shrinkage parameter $k$ is of interest.

One must note that solving the optimization problem (1.13) is not the only way of yielding the RRE. It can also be obtained by solving a RSS of another data, say augmented data. To be specific, consider the following augmentation approach. Let
$$\boldsymbol{X}^{}=\left[\begin{array}{c} \boldsymbol{X} \ \sqrt{k} \boldsymbol{I}{p} \end{array}\right], \quad \boldsymbol{Y}^{}=\left[\begin{array}{c} \boldsymbol{Y} \ \mathbf{0} \end{array}\right]$$
Assume the following multiple linear model,
$$Y^{}=X^{} \beta+\epsilon^{},$$ where $\boldsymbol{e}^{}$ is an $(n+p)$-vector of i.i.d. random variables. Then, the LSE of $\boldsymbol{\beta}$ is obtained as
\begin{aligned} \boldsymbol{\beta}{n}^{} &=\min {\boldsymbol{\beta}}\left{\left(\boldsymbol{Y}^{}-\boldsymbol{X}^{} \boldsymbol{\beta}\right)^{\top}\left(\boldsymbol{Y}^{}-\boldsymbol{X}^{} \boldsymbol{\beta}\right)\right} \ &=\left(\boldsymbol{X}^{} \boldsymbol{X}^{}\right)^{-1} \boldsymbol{X}^{ \top} \boldsymbol{Y}^{*} \ &=\left(\boldsymbol{X}^{\top} \boldsymbol{X}+k \boldsymbol{I}{p}\right)^{-1} \boldsymbol{X}^{\top} \boldsymbol{Y} \ &=\hat{\boldsymbol{\beta}}_{n}^{\mathrm{RR}}(k) . \end{aligned}
Thus, the LSE of the augmented data is indeed the RRE of the normal data.

## 统计代写|似然估计作业代写Probability and Estimation代考| Introduction to Ridge Regression

∂小号(b)∂b=−2X⊤是+2X⊤Xb=0

$$\tilde{\boldsymbol{\beta}} {n}=\left(\boldsymbol{X}^{\top} \boldsymbol{X}\right)^{-1} \boldsymbol{X}^{ \top} \boldsymbol{Y} \text {. }$$ 假设和(e)=0和 $\mathbb{E}\left(\boldsymbol{\varepsilon} \boldsymbol{e}^{\top}\right)=\sigma^{2} \boldsymbol{I} {n}F这rs这米和\sigma^{2} \in \mathbb{R}^{+}.吨H和n,吨H和在一种r一世一种nC和−C这在一种r一世一种nC和米一种吨r一世X这F大号小号和一世sG一世在和nb是曾是⁡(b~n)=σ2(X⊤X)−1ñ这在,在和C这ns一世d和r吨H和C一种n这n一世C一种lF这r米这F吨H和米在l吨一世pl和l一世n和一种rr和Gr和ss一世这n米这d和l吨这一世ll在s吨r一种吨和H这在l一种rG和和一世G和n在一种l在和s这F吨H和d和s一世Gn米一种吨r一世X\boldsymbol{X}^{\top} \boldsymbol{X}$ 可能会影响估计的效率。

X~n=(吨⊤吨−1吨⊤是 =Λ−1吨⊤是的方差-协方差矩阵X~n是（谁）给的

## 统计代写|似然估计作业代写Probability and Estimation代考| Ridge Regression Estimator: Ridge Notion

\min {\beta}\left{(YX \boldsymbol{\beta})^{\top}(\boldsymbol{Y}-\boldsymbol{X} \boldsymbol{\beta})\right} \quad \text {这样 } \sum{j=1}^{p} \beta_{j}^{2} \leq t\min {\beta}\left{(YX \boldsymbol{\beta})^{\top}(\boldsymbol{Y}-\boldsymbol{X} \boldsymbol{\beta})\right} \quad \text {这样 } \sum{j=1}^{p} \beta_{j}^{2} \leq t

(是−Xb)⊤(是−Xb)+ķ|b|2,|b|2=∑j=1pbj2

\min {\boldsymbol{\beta}}\left{(\boldsymbol{Y}-\boldsymbol{X} \boldsymbol{\beta})^{\top}(\boldsymbol{Y}-\boldsymbol{X} \ boldsymbol{\beta})+k|\boldsymbol{\beta}|^{2}\right}=\min {\boldsymbol{\beta}}{\operatorname{PS}(\boldsymbol{\beta})}，\min {\boldsymbol{\beta}}\left{(\boldsymbol{Y}-\boldsymbol{X} \boldsymbol{\beta})^{\top}(\boldsymbol{Y}-\boldsymbol{X} \ boldsymbol{\beta})+k|\boldsymbol{\beta}|^{2}\right}=\min {\boldsymbol{\beta}}{\operatorname{PS}(\boldsymbol{\beta})}，

∂磷小号(b)∂b=−2X⊤是+2X⊤Xb+2ķb=0

b^nRR(ķ)=(X⊤X+ķ一世p)−1X⊤是.

X=[X ķ一世p],是=[是 0]

\begin{对齐} \boldsymbol{\beta}{n}^{} &=\min {\boldsymbol{\beta}}\left{\left(\boldsymbol{Y}^{}-\boldsymbol{X}^ {} \boldsymbol{\beta}\right)^{\top}\left(\boldsymbol{Y}^{}-\boldsymbol{X}^{} \boldsymbol{\beta}\right)\right} \ & =\left(\boldsymbol{X}^{} \boldsymbol{X}^{}\right)^{-1} \boldsymbol{X}^{ \top} \boldsymbol{Y}^{*} \ &= \left(\boldsymbol{X}^{\top} \boldsymbol{X}+k \boldsymbol{I}{p}\right)^{-1} \boldsymbol{X}^{\top} \boldsymbol{Y } \ &=\hat{\boldsymbol{\beta}}_{n}^{\mathrm{RR}}(k) 。\end{对齐}\begin{对齐} \boldsymbol{\beta}{n}^{} &=\min {\boldsymbol{\beta}}\left{\left(\boldsymbol{Y}^{}-\boldsymbol{X}^ {} \boldsymbol{\beta}\right)^{\top}\left(\boldsymbol{Y}^{}-\boldsymbol{X}^{} \boldsymbol{\beta}\right)\right} \ & =\left(\boldsymbol{X}^{} \boldsymbol{X}^{}\right)^{-1} \boldsymbol{X}^{ \top} \boldsymbol{Y}^{*} \ &= \left(\boldsymbol{X}^{\top} \boldsymbol{X}+k \boldsymbol{I}{p}\right)^{-1} \boldsymbol{X}^{\top} \boldsymbol{Y } \ &=\hat{\boldsymbol{\beta}}_{n}^{\mathrm{RR}}(k) 。\end{对齐}

## 有限元方法代写

tatistics-lab作为专业的留学生服务机构，多年来已为美国、英国、加拿大、澳洲等留学热门地的学生提供专业的学术服务，包括但不限于Essay代写，Assignment代写，Dissertation代写，Report代写，小组作业代写，Proposal代写，Paper代写，Presentation代写，计算机作业代写，论文修改和润色，网课代做，exam代考等等。写作范围涵盖高中，本科，研究生等海外留学全阶段，辐射金融，经济学，会计学，审计学，管理学等全球99%专业科目。写作团队既有专业英语母语作者，也有海外名校硕博留学生，每位写作老师都拥有过硬的语言能力，专业的学科背景和学术写作经验。我们承诺100%原创，100%专业，100%准时，100%满意。

## MATLAB代写

MATLAB 是一种用于技术计算的高性能语言。它将计算、可视化和编程集成在一个易于使用的环境中，其中问题和解决方案以熟悉的数学符号表示。典型用途包括：数学和计算算法开发建模、仿真和原型制作数据分析、探索和可视化科学和工程图形应用程序开发，包括图形用户界面构建MATLAB 是一个交互式系统，其基本数据元素是一个不需要维度的数组。这使您可以解决许多技术计算问题，尤其是那些具有矩阵和向量公式的问题，而只需用 C 或 Fortran 等标量非交互式语言编写程序所需的时间的一小部分。MATLAB 名称代表矩阵实验室。MATLAB 最初的编写目的是提供对由 LINPACK 和 EISPACK 项目开发的矩阵软件的轻松访问，这两个项目共同代表了矩阵计算软件的最新技术。MATLAB 经过多年的发展，得到了许多用户的投入。在大学环境中，它是数学、工程和科学入门和高级课程的标准教学工具。在工业领域，MATLAB 是高效研究、开发和分析的首选工具。MATLAB 具有一系列称为工具箱的特定于应用程序的解决方案。对于大多数 MATLAB 用户来说非常重要，工具箱允许您学习应用专业技术。工具箱是 MATLAB 函数（M 文件）的综合集合，可扩展 MATLAB 环境以解决特定类别的问题。可用工具箱的领域包括信号处理、控制系统、神经网络、模糊逻辑、小波、仿真等。