Linear statistical method (I)

Statistical Method (I) 統計方法 (I)

Instructor: Pi-Wen Tsai 蔡碧紋

Office: M 205

Phone: 7734-6615

Email: pwtsai@math.ntnu.edu.tw

Lectures: 星期二上午 9-12

Office hours:星期二下午2-4 or by email appointment

Course Webpage: http://math.ntnu.edu.tw/~pwtsai/SM98/index.html

最新消息：

學期成績：(已移除) final.html

請列印 ch9.pdf 12/22 去 M411 上課
1/5, 1/12, 1/13 報告請準備投影片 (儘量為 pdf 檔)
每人約 15 分

作業

HW			R.
1(ch2+ch3)	ch2	4,17,22,25	ch2.r ch2.pdf
	ch3	5,6,21,22,33	ch3.R ch3.pdf ch301.pdf
			hw1.r
2	ch4	4, 17, 18	ch4.R ch4.pdf
	ch5	modified 2(pdf), 12, 13	ch5.pdf ch5.R
		資料檔4.18的資料有錯誤,請修正x1第二行的-1為 1 ex4-18.txt	ex-5-1.csv, ex-5-2.csv, blood.txt
3	Ch6	3, 10, 16 (data: tb15.txt) read.table("tb15.txt", header=T, sep="\t")	ch6.pdf ch6.R
	ch7	11, 14, 19, 20	ch7.pdf
4	ch8	13, 14	ch8.pdf ch8.R tool.txt
	ch9	7, 21, 22, 23.	wine.pdf wine.R (model checking)
			ch9.pdf ch9.R ex-9-1.txt
			cv_boot.pdf cv_boot.R

課程目標：

統計的應用不勝枚舉，只要牽涉到資料的蒐集、整理與分析，就必須有統計的基礎，才能得到合理的結果。本課程以統計學為基礎，進一步探討統計之理論及應用。目的在於使學生了解統計的應用方向、基本理論與資料分析技巧。本課程設計給已修過機率與統計學的學生，課程主要範圍包括迴歸模式分析、變異數分析等之理論及應用。

關鍵字: 估計及假設檢定、迴歸模式分析、變異數分析

先修課程: 機率、統計學、線性代數。

Text Books:

D.C.Montgomery, E.A. Peck, and G.G. Vining. Introduction to Linear Regression Analysis 4th edition, John Wiley & Sons, New York, 2007. (歐亞書局代理)

Reference Books:

SimpleR -- Using R for the Introductory of Statistics.「直接下載]
Kutner, M. H, Nachtsheim, C. J., Neter, j. and Li. W Applied Linear Statistical Models 5th ed.(華泰書局代理)

Syllabus:

Regression Analysis: Quantitative explanatory variables.

簡單線性迴歸模式: Simple linear regression model

least squares estimation.
geometry of least squares.
maximum likelihood estimation.
analysis of variance, R2, Test for goodness, lack of fit test.

多變量線性迴歸模式: Multiple regression models

General linear model
Additional Sum of squares: ANOVA
Hypothesis testing
Goodness of fit test
Multicollinearity

適度性檢定: Accessing model adequacy

Residual analysis
Lack of fit test

Transformations and Weighted LS
1. Variance-stabilizing transformations
2. Box-Cox transformation
3. Generalized and Weighted LS

模型診斷: Regression Diagnostics:

Outliers, leverage and Influence measure

Polynomial regression

Indicator Variables: Qualitative explanatory variables:

ANOVA (Analysis of variance),
ANCOVA(Analysis of covariance)

統計建模: Model building and variable selection

All possible regressions
Automatic methods

Special topics (time permitting)

Multicollinearity
The problem of Corelated errors
Robust regression
Generalized Linear models

Evaluation Details:

Assignments: There will be regular assignments worth 25% of the total mark. Assignment will require both analytic work, as well as the use of statistical software.

Exams: The two exams worth 60% of the total mark. They will take place during the regularly scheduled class time.

Project: There will be a project worth 15% of the total mark. The project will require a written report and a presentation.
可自己分析資料或說明比較已有的文獻分析.

Download R (exe files) to your own personal computer.

Open the Internet Explorer and go to the website http://cran.r-project.org
Click on the link Windows (95 and later) in the top box.
Click on the link base.
Click on the link R-2.9.2-win32.exe You should save this file to your desktop.
After the file has downloaded, double-click on the icon on your desktop and follow the instructions to install the program.
The installation should place a shortcut on your desktop. You simply double-click on the shortcut to run R.

----
A platform-independent basic-statistics GUI (graphical user interface) for R: Rcmdr
A useful Editor for R on Windows: Tinn-R.

Source in UCLA: Statistical computing
R-regessin-reference card: by R. Andersen ref.pdf
Vito Ricci's R regression reference card: ricci.pdf