Spring 2013


Course Info

  • Syllabus
  • Instructor: Yu-Ling Tseng.      Office Hours: W. 12:00 ~ 13:00  (理A409)
  • Prerequisites: Probability Theory, Statistics, Linear Algebra

  • R Sample Programs ( in text format unless specified otherwise)  :  You will need to use R for most assignments (find and use the respective data set for each problem in the CD provided with your textbook) & you should find these sample programs quite useful.

    Programs (file name gives the date of its use in class)

    SetUp(1) shows you how to:
    1. devide graphical windows
    2. assign variables/sequences
    3. plot and make titles for the graph
    4. generate normally distributed observations/make histograms
    5. do simple regressions and related plots  (you should see the effect of sample size and s.d.  on the fitted reg. lines)
    6. quit R
    SetUp(2) shows you how to:
    1. scan a data set in your working directory into R
    2. get figures and tables in the textbook with the Toluca Company examples
    3. construct confidence intervals for reg. coeff.'s
    4. quit R, again.
    ConfInt ConfInt shows you ,with the Toluca Company examples, how to
    3. construct confidence intervals for reg. coeff.'s, confidence/prediction  intervals for the mean response at given predictor x's values
    4. construt conf. band for the entire reg line,
    5. overlay three type of intervals in one plot for better comparison
    6. save the output to a file for preparing your homework, or save the R commands for later use
    MLR1This program shows you how to
    1. some basic matrix operations in R,
    2. how to obtain the design matrix after fitting a simple linear regression model,

    3. do multiple linear regression
    4. get figures and tables in the textbook with the Dwaine Studios examples
    5. make basic scatter plots for M-L-R data analysis
    6. obtain the design matrix
    MLR2  shows you ,with Awaine Studios example again, how to
    get (simultaneous) conf. intervals for reg. coefficients,  confidence/prediction  intervals for the mean response at given predictors's values
    MLR3 shows you how to
    1. obtain SSEs from the full model and the reduced model
    2. obtain F(1-alpha; m, k) in R
    3. use the general linear test approach  for testing certain hypotheses (by giving 3 examples)
    ResidPlotThis program ResidPlot
    1.  let you get a feeling as how a random sample of size n from N(0, 1) would look like in time sequence plots, and in histograms;
    2. with simulated regression data.....  shows you some basic residual-plots for diagnostic in a regression analysis
    Please note that how a violation of certain assumptions made in reg model affect the display........
                                WLSEThis program shows you how to do W.L.S.E. when non-constant variances occurs......
    esp. shows you how to get  figures and tables in the textbook with Blood Pressure Example on p427 .
    varstabtransThis program varstabtrans illustrates a complete process when analyzing a real data set with nonconstant variance problem........
    Instead on using W.L.S.E. (which is covered in WLSE), we try transformations when nonconstant variances occur in this program.
    We run through the Case Example -- Plutonium Measurement on p 141 of textbook.
    Esp. you learn how to delete some data points from a data set, how to update model , and how to get basic diaqnostics residual plots.

    R ClassRoom
    SimpleR (by John Verzani)
    Data Sets of KNNL (5thED, .txt format)
    Yes 公佈欄:  You may find useful R programs  here:

    Topics for presentations:   0603 或 0606 報告。 各組報告時間 35~45 分鐘

    1. 7.1 Extra sum of squares, p. 256 ~ p. 262

    7.2 Use of extra sum of squares in tests for regression coefficients, ~ p. 265


    2. 7.6 Multicollinearity and its effects, p. 278 ~ p. 289

      11.2 Multicollinaerity remedial measures-Ridge regression, p. 431~436

    組員 : 陳思卉.林孟緯.黃莉雅

    3. 8.1~ 8.2 Polynomial regression models, Interaction regression models, p. 294~ 312

    組員: 阿兩  房雨蓉  肥羊

    4. 8.3~8.4 Qualitative predictors, Some considerations in using indicator variables,

    p. 313~323


    5. 8.5~8.7 Modeling interactions between quantitative and qualitative predictors, More complex models, Comparison of two or more regression functions, p. 324~342

    6. 9.1~9.3 On model selection (I), p. 343~360

    7. 9.4~9.6 On model selection (II), p. 361~383

    Assignments (No Need to Hand in!)                                                                                                           
    0523Ch.6 : 5 (a, b), 6 (a, b, c), 10 (a), 11 (a, b, c), 15 (c), 16 (a, b, c,),  17, 19
    Ch.3 : 3 (a, b, c, d.) (For d, only need to prepare a normal probability plot, i.e. the Q-Q plot)
             4 (a, b, c, d, e, f, h) (For e, only need to prepare a normal probability plot, i.e. the Q-Q plot)
            6 (a, b, c.) ( For c, only need the Q-Q plot)
            8 (a, b, c, d). ( Only Q-Q plot for d)

    0418Ch.2 : 3, 4, 8 (a, c), 10, 13, 16, 17, 23 (a, c), 30
    Ch.6 : 2, 4, 7, 22, 23, 24, 25, 26
    0415problems given in 0415's class 
    0411Ch.1 : 45
    Ch.5 : 17, 18, 19
    0307Ch.1 :  8, 20, 27, 30,33, 34, 35, 36, 39 (a) , 40, 41


    1. R website (original, mirror @ NTU)        
    2. R ClassRoom
    3. SimpleR (by John Verzani)
    4. Document Reader: Ghostview, Acrobat Reader


    Last modified: 060912