To Home Page

CSC423 -- Examples

Example Zipfiles

• day1.zip   day2.zip   day3.zip   day4.zip
  day5.zip   day6.zip   day7.zip   day9.zip

Index of Examples

• Univariate Analysis: ExamSco  NIST  BloodPressure  Subset  Reformat
• One and two-sample t-tests: Bones  CO  ReadingPaired  ReadingIndep
• Correlation: BearsCorrSimple  BearsCorr  TireWear
• Regression: SpringReg  BearsReg  Hamster  CPUReg1
• Matrix Form of Regression Model: MatrixSpring  CrudeOil
• Backward and All Subsets Selection:  CPUReg2   CPUReg3
• Confidence and Prediction Intervals: Hamster
• Influence Points and Multicollinarity: InfluenceSimulation   Sales
• Diagnostic Plots: Diagnostics
• Dummy Variables: WLoss1   WLoss2   WLoss3   Movies
• Interactions and Polynomial Models: GFClocks   Olympic
• The Durbin-Watson Test for Independence: SalesDW
• Graphing: SGScatter   G3D  
• Stepwise Selection and Model Validation: PropVal1   PropVal2   Prediction   PropVal3   PropVal4   GLMSelect   Crime  
• Transforms: Pendul  BodyBrain
• Maximum of 2D Surface: Surface 
• ANOVA Models: WLoss4   Orange   Orange2
• Logistic Regression: Discrim   Depress   Remission
• χ² Goodness-of-Fit Test: FairCoin1   FairCoin2   FairCoin3
• χ² Test of Independence: Cold1   Cold2   Cold3
• Case Studies: Condo   Trucking

Univariate Analysis

• ExamSco Example -- Hypothetical Exam Scores
   
  Create a dataset of hypothetical exam scores.
  Print the data, and calculate descriptive statistics.
   
  

  Data Files	SAS Source Code	R Source Code
  	examsco1.sas SAS Output SAS Log File	examsco1.R R Output
  examsco2.txt	examsco2.sas SAS Output	examsco2.R R Output
  examsco3.txt	examsco3.sas SAS Output	examsco3.R R Output

  
   
• NIST Example -- NIST Prototype Weight Weighings
   
  100 weighings of a 10g prototype weight, obtained at the National Institute of Standards and Technology (formerly the National Bureau of Standards).
  Calculate these statistics for the dataset: mean, standard deviation, median, IQR, percentiles 5, 10, 90, 95.
  Create these plots: histogram, boxplot, normal plot.
   
  

  Data File	SAS Source Code	R Source Code
  nist.txt	nist.sas SAS Output Histogram Normal Plot Boxplot	nist.R R Output Graphics Output

   
• Blood Pressure Example -- Univariate Statistics and Side-by-side Boxplots
   
  The blood pressure of fourteen persons is measured, both in standing and supine position. The blood pressure readings are recorded in the datasets blood-pressure1.txt and blood-pressure2.txt. Here are the SAS and R scripts with their output that perform the tasks in the Lab 1 Description.  You must create your own bp1cleaned.txt and bp2cleaned.txt files with outliers removed and place them in the datasets folder.
   
  

  Data File	SAS Source Code	R Source Code
  blood-pressure1.txt blood-pressure2.txt	blood-pressure.sas SAS Output boxplots.png hist1-supine.png hist2-supine.png hist3-supine.png normplot-standing.png normplot-supine.png	blood-pressure.R R Output bp.pdf

  
  
• Subset Example -- Subset a Dataset (SAS) or Data Frame (R)
   
  Create the dataset or data frame named all. Then select only the rows with gender=F and the columns name, midterm, and final to create the new dataset named women.
   
  

  Data File	SAS Source Code	R Source Code
  examsco2.txt	subset.sas SAS Output	subset.R R Output

  
  
• Reformat Example -- Reformat Dataset for Side-by-side Boxplots
   
  The each pair in the reading-paired.txt dataset consists of two students with similar IQs, one taught with a standard reading method and one taught with the new reading method that is being investigated. Reformat the data to put all data in one variable with a second categorical variable with the values std and new to differentiate between the two teaching methods. Then create side-by-side boxplots of the data.
   
  

  Data File	SAS Source Code	R Source Code
  reading-paired.txt	reformat.sas SAS Output sidebysidebp.gif	reformat.R R Output sidebysidebf.pdf

• Bones Example -- One-sample t-test
   
  From Mendenhall and Sincich, p. 45: Humerous bones from the same species of animal have approximately the same length-to-width ratios. It is known that species A has a mean ratio of 8.5. Suppose that 41 fossil humerous bones were unearthed at a site where species A is known to have flourished. (We assume that all bones are from the same species.) Perform a level 0.05 one-sample t-test to test the null hypothesis that the difference of the average length-to-width ratio from the mean ration 8.5 of species A is just chance variation. Note: because n > 30, a z-test could be performed; we perform a t-test here because statistical software does not distinguish between z-tests and t-tests.
   
  

  Data File	SAS Source Code	R Source Code
  bones.txt	bones.sas SAS Output	bones.R R Output

  
  
• CO Example -- One-sample t-test
   
  The carbon monoxide (CO) level in a manufacturing plant is supposed to be about 50 parts per million (ppm). However the actual CO levels are quite variable. Five CO measurements are taken at various times during the day:
   
  
  58    63    48    52    68
  
  Test the null hypotheses that μ = 50 ppm for the CO concentration at the manufacturing plant.
   
  

  SAS Source Code	S Source Code
  co.sas SAS Output	co.R Output

  
  
• ReadingPaired Example -- Paired-sample t-test
   
  From Mendenhall and Sincich, p. 57: To measure whether a new method for teaching reading works better than the standard method, eight pairs of children that have had difficulties learning how to read are located. Each pair is arranged to contain two children with similar IQ scores. Then one child out of each pair is taught using the standard teaching method (designated as Group 1), the other child out of the pair, is taught using the new teaching method (designated as Group 2). The difference score for each pair is computed: d_i = x_2i - d_1i. Perform a paired-sample t-test to test the null hypothesis that the difference between the teaching methods is just chance variation.
   
  

  Data File	SAS Source Code	R Source Code
  reading-paired.txt	reading-paired.sas SAS Output	reading-paired.R R Output

  
  
• ReadingIndep Example -- Independent Two-sample t-test
   
  Mendenhall and Sincich: p. 55: To measure whether a new method for teaching reading works better than the standard method, 22 children that have had difficulties learning how to read are selected. 10 children are are taught reading using the new method; 12 children are taught using the standard method. All children are then given a standardized test. Perform a test of the null hypothesis that there is no difference between the two teaching methods.
   
  

  Data File	SAS Source Code	R Source Code
  reading-indep.txt	reading-indep.sas SAS Output	reading-indep.R R Output

• BearsCorrSimple Example -- Covariance, and Correlation, and Scatterplot
   
  Using the 1985 Chicago Bears roster (the year after they won the Super Bowl), compute sample mean and SD of height and weight, sample covariance and correlation of height and weight. Also show the scatterplot of weight vs. height.
   
  

  Data File	SAS Source Code	R Source Code
  bears85.txt	bears.sas SAS Output bears-simple.png	bears.R R Output bears-simple.pdf

  
  
• BearsCorr Example -- Scatterplots, Covariances, and Correlations
   
  Perform various procedures on height and weight data from the 1985 Chicago Bears roster. Include scatterplots and compute the covariances and correlations.
   
  

  Data File	SAS Source Code	R Source Code
  bears85.txt	bears.sas SAS Output bears1.gif bears2.gif	bears.R R Output bears.pdf

  
  
• TireWear Example -- Nonlinear Association Makes Sample Correlation Misleading
   
  Look at the scatterplot and correlation of tire mileage vs. time pressure.
   
  

  Data File	SAS Source Code	R Source Code
  tirewear.txt	tirewear.sas SAS Output tirewear.gif	tirewear.R R Output tirewear.pdf

• SpringReg Example -- Regression Models, and Scatter Plots
   
  Determine the regression through the origin and simple linear regression models. Include a scatterplot with regression line for each model.
   
  

  SAS Source Code	R Source Code
  springreg.sas SAS Output springreg1.gif springreg2.gif	springreg.R R Output springreg.pdf

  
  
• BearsReg Example -- Regression Model, Residual Analysis
   
  Find the simple linear regression equation for predicting weight from height for the 1985 Chicago Bears Dataset. Include a scatterplot and plots for regression diagnostics.
   
  

  Data File	SAS Source Code	R Source Code
  bears85.txt	bearsreg.sas SAS Output Scatterplot Residuals vs. Height Resiguals vs. Predicted Normal Plot	bears.R R Output bearsreg.pdf

  
  
• Hampster Example -- Regression Model, Residual Analysis
   
  Predict the lifetime of a hamster from its percent of life hibernating. Create the scatterplot, the residual plot, and normal plot of residuals.
   
  

  Data File	SAS Source Code	R Source Code
  hamster.txt	hamster.sas SAS Output Scatterplot Regression Plot Normal Plot	hamster.R R Output Graphs

  
  
• CPUReg1 Example -- Regression Model, Residual Analysis
   
  Predict the amount of CPU time from number of lines of code. Use the simple linear regression model y = a + bx. We will use the other independent independent variables later for a multiple regression model.
   
  

  Data File	SAS Source Code	R Source Code
  cpu.txt	cpureg.sas SAS Output Scatterplot Residual Plot Normal Plot	cpureg1.R R Output R Plots

• MatrixSpring Example -- Matrix Form of Regression Model
   
  Formulate both the regression through the origin and the simple linear regression model in the matrix form y = Xβ + ε. Obtain the estimated regression parameters and their standard errors.
   
  

  SAS Source Code	R Source Code
  spring-matrix.sas SAS Output	R Script and Output for Model 1 R Script and Output for Model 2

  
  
• CrudeOil -- Matrix Form of Regression Model
   
  This example uses the Crude Oil dataset crude-oil.txt described in The Crude Oil Example description. The LSE parameters are obtained in with SAS and R using proc reg and the lm function, respectively. In addition, the estimated parameters are obtained from the regression model y = Xβ + ε using β^{^} = (X^TX)^-1 X^TY.
   
  

  SAS Source Code and Output	R Source Code and Output
  crude-oil.sas SAS Output	crude-oil.R R Output

Backward and All Subsets Selection

• CPUReg2 Example -- Backward Selection
   
  Reference: Montgomery and Peck, Introduction to Linear Regression Analysis, Wiley Interscience, 1982.
   
  Start with the complete model (all independent variables included), then eliminate non-significant variables one at a time.

  Data File	SAS Source Code	R Source Code
  cpu.txt	cpureg2.sas SAS Output r. * p. r. * linesout r. * steps r. * devices r. * nqq.	cpureg2.R R Output R Plots

  
  
• CPUReg3 Example -- All Subsets Selection
   
  Compute R-squared and adjusted R-squared for all possible subsets.

  Data File	SAS Source Code	R Source Code
  cpu.txt	cpureg3.sas SAS Output	cpureg3.R R Output

• Hamster Example -- Prediction and Confidence Intervals
   
  Obtain the X^T X, X^T y, (X^T X)^-1 matrices. Also graph the 95% confidence interval for y^ and the 95% prediction interval for a new observation.
   
  

  Data File	SAS Source Code	R Source Code
  hamster.txt	hamster.sas SAS Output conf-pred.gif	hamster-matrix.R R Output (Matrix) hamster-limits.R R Output (Limits) R Plot

• InfluenceSimulation Example
   
  Create a dataset of 50 points with x and y independent N(0, 1) random variables. (The data sets are randomly generated, so they are different in SAS than they are in R. Now add (x, y) = (20, 20). This is Model 1. The point (x, y) is seen to be an influence point: it has a large influence on the slope. See SAS Plot 1 and the first plot in R Plots.  Now add (x, y) = (0, 20). (x, y) is still an influence point but in a different way: it has a large influence on the intercept. See SAS Plot 2 and the second plot in R Plots.
   
  

  SAS Source Code	R Source Code
  inf-sim.sas SAS Output SAS Plot 1 SAS Plot 2	inf-sim.R R Output R Plots

   
• Sales Example
   
  Multicollinarity means that at least one of the independent variables is strongly correlated with a linear combination of some of the other independent variables.

  Data File	SAS Source Code	R Source Code
  sales.txt	sales.sas SAS Output	sales.R R Output

• Diagnostics Example
   
  Both SAS and R are capable of producing a suite of regression diagnostic plots without explicitly constructing them. With SAS, this is accomplished by using ODS (Output Delivery System). With R, pass the model object from the lm function to the plot function.
   
  

  Dataset	SAS Source Code	R Source Code
  cpu.txt	diagnostic.sas SAS Output SAS ODS Plots	diagnostic.R R Output R Plots

• WLoss1 Example -- ANOVA and Dummy Variables
   
  Perform a classical analysis of variance (ANOVA) test for testing whether the weight loss center makes a difference in the amount of weight lost.
   
  

  SAS Source Code	R Source Code
  wloss1.sas SAS Output r. * p. r. * nqq.	wloss1.R R Output R Graphs

  
  
• WLoss2 Example -- ANOVA and Dummy Variables
   
  Perform a classical analysis of variance (ANOVA) test for testing whether the weight loss center makes a difference in the amount of weight lost. There are five weight loss centers. Use the Bonferroni method for post-hoc comparisons to see which centers are different from which other centers.
   
  

  SAS Source Code	R Source Code
  wloss2.sas SAS Output r. * p. r. * nqq.	wloss2.R R Output R Graphs

  
  
• WLoss3 Example -- Dummy Variables
   
  A simplified version of the WLoss1 and WLoss2 Examples. Set up a regression model for predicting the weight loss based on the dummy variable of the weight loss center (A, B, C, D, or E). There is no dummy variable for Center E.
   
  

  Data File	SAS Source Code	R Source Code
  wloss.txt	wloss3.sas SAS Output r. * p. r. * nqq.	wloss3.R R Output R Graphs

  
  
• Movies Example
   
  Pick best regression model, then show scatterplots and residual plots using plotting symbols 0 and 1, depending on the values of star_dum or sum_sum. We can't do this with the plot option of proc reg, so instead, save the residual values and plot them using proc gplot.
   
  Use these encodings for isstar (is there a big-name star in the movie?): NoStar => 0, Star => 1. Use these encodings for issummer (movie released in the summer): NoSummer => 0, Summer => 1.
   
  

  Data File	SAS Source Code	R Source Code
  movies.txt movies2.txt	movies.sas SAS Output opening * budget = star_dum opening * budget = sum_dum residuals * predicted = star_dum residuals * budget = star_dum residuals * sum_dum	movies.R R Output R Graphs

• GClocks Example -- A Model with an Interaction Term
   
  When selling grandfather clocks at aution, does the age of the clock interact with the number of bidders?
   
  

  Data File	SAS Source Code	R Source Code
  gfclocks.txt	gfclocks.sas SAS Output Residual Plot 1 Normal Plot 1 Residual Plot 2 Contour Fit Plot	gfclocks.R R Output R Plots

• SalesDW Example -- The Durbin-Watson Test for Independence
   
  Check sales data for independence of residuals
   
  

  Data File	SAS Source Code	R Source Code
  sales.txt	salesdw.sas SAS Output SAS Residual Plot	salesdw.R R Output R Residual Plot

• SGScatter Example
   
  Graph multidimensional data as a scatterplot matrix.
  

  Data File	SAS Source Code	R Source Code
  iris.txt	sgscatter.sas SAS Output SAS Output Graph	sgscatter.R R Output R Output Graph

  
  
• G3D Example
   
  Graph multidimensional data as a 3-d plot.
  

  Data File	SAS Source Code	R Source Code
  iris.txt	g3d.sas SAS Output Graph	g3d.R R Output Graph

• PropVal1 Example -- All Possible Regression Models
   
  Compute regression model statistics for all possible subsets of independent values. Use the SAS proc rsquare.
   
  

  Data File	SAS Source Code
  propval.txt	propval1.sas SAS Output

  
  
• PropVal2 Example -- Forward, Backward, and Stepwise Selection
   
  Use forward, backward, and stepwise regression work for model selection.
   
  

  Data File	SAS Source Code	R Source Code
  propval.txt	propval2.sas SAS Output	propval2.R R Output

  
  
• Prediction Example -- New Data Points
   
  Compute the predicted values and prediction intervals of new data points.
   
  

  SAS Source Code	R Source Code
  prediction.sas SAS Output	prediction.R R Output

  
  
• PropVal3 Example -- Holdout Sample
   
  Select and use a holdout sample to validate a regression model.
   
  

  Data File	SAS Source Code	R Source Code
  propval.txt	propval3.sas SAS Output	propval3.R R Output

  
  
• PropVal4 Example -- Small Test Set
   
  Select and use a small test set to validate a regression model. This example is similar to PropVal3, but with steps executed by hand to make it easier to understand the code. The R script in PropVal3 is easier to understand than the SAS script in PropVal3, so R scripts are not provided for this example.
   
  

  Data File	SAS Source Code
  propval.txt training-test.txt actual-predicted-mean.txt	split.sas SAS Output from split.sas computepredicted.sas SAS Output from computepredicted.sas crossval.sas SAS Output from crossval.sas

  
  Directions:
   
  
  1. Run the SAS script split.sas.
      
  2. Copy the output from Step 1 to the new input file test-training.txt. Remove Page 1. On Page 2, move all lines with Selected=0 to the end of the file. Save the y values with Selected=0 to put in the first column of the input file actual-predicted-mean.txt, but in the training-text.txt file, change all y values to missing ( . ). Remove all blank lines at the bottom of training-test.txt. Also remove the title, page number, column headers, and new line character at the top of each page.  You can leave the column headers at the top of page 1 if you wish.
      
  3. Run crossval.sas.
      
  4. Go back to the input file actual-predicted-mean.txt. (You already copied the actual y values from the split.sas output into this file.) Now copy the predicted values with Dependent Value = . into actual-predicted-mean.txt as Column 2. Copy the mean value of y on Page 3 of the actual-predicted-mean.sas output into each row of Column 3.
      
  5. Run crossval.sas.
      
  6. Compute the crossvalidation R-squared as R_cv² = SSE / SST, where SSE is computed as the sum of the ressq values in the crossval.sas output and SST is computed as the sum of the devsq values in the crossval.sas output. Looking at the proc means output in the crossval.out output, we have
      
     
     R_cv² = 1 - SSE / SST = 1 - 194.0473191 / 183.2978207 = -0.05864499.
     
     This shows that the regression model is not good. We have too few observations relative to the number of independent variables. A good value of R_cv² should be greater than 0.20, or at least close to the R² value obtained by the training set model.
  
  
• GLMSelect Example -- Crossvalidation
   
  Use crossvalidation to choose best variables.
   
  

  Data File	SAS Source Code
  propval.txt	glmselect.sas SAS Output

  
  
• Crime Example -- All Subsets Regression
   
  Compute best regression model using SAS proc rsquare or R leaps function. Install the leaps package before running the R example.
   
  

  Data File	SAS Source Code	R Source Code
  crime.txt	crime.sas SAS Output	crime.R R Output

• Pendul Example -- Regression Model, Residual Analysis
   
  Predict the period of a pendulum from its length using four regression models:
   
  
  1. Simple Linear Regression
      
  2. Quadratic Regression
      
  3. Using a Square Root Transformation
      
  4. Using a Log-log transformation
  
  This example is interesting because we know the correct answer from physics:
   
  
  period = 2 π √length / g, where π = 3.14159265 and g = 1.97078 m/sec²
  
  

  Data File	SAS Source Code	R Source Code
  pendul.txt	pendul.sas SAS Output SAS plots not shown, see R plots	pendul.R R Output R Plots

  
  
• BodyBrain Example -- Log-log Transformation
   
  Because the body and brain weight of animals vary by many orders of magnitude, use a log transformation on both the independent and dependent variables before obtaining the regression model

  Data File	SAS Source Code	R Source Code
  body-brain.txt	body-brain.sas SAS Output SAS plots not shown, see R plots	body-brain.R R Output R Plots

• Surface Example
   
  Find maximum point of surface by setting the partial derivatives to 0 and solving the system of 2 equations in 2 unknowns.
   
  

  Solve by Hand	Solve with SAS	Solve with R
  Solve by Hand	solve.sas	solve.R 3D Surface and Contour R Plots

  
  Here is the R script that produces the 3D Surface and Contour R Plots shown in the PDF file with link above. The SAS proc glm produces the contour plot automatically, but you need to call it like this, without defining your own quadratic and interaction terms:
  

  proc glm;
     model t t*t p p*p p*t;
  

  If you use your own quadratic and interaction terms, proc glm does not know that you are using a full quadratic model.

ANOVA Models

• WLoss4 Example -- One-way ANOVA Model
   
  Perform a one-way ANOVA test on the the weight loss data using proc anova and glm.  Use the lm, aov, and ?? R functions. In both the SAS and R examples, the factor levels are entered directly and the dummy variables are automatically generated.
   
  

  Data File	SAS Source Code	R Source Code
  wloss4.txt	wloss4.sas SAS Output Boxplots Residual Plot Normal Plot	wloss4.R R Output R Graphs

  
  
• Orange Example -- Two-way ANOVA model with and without Interaction
   
  Investigate the effect of tree variety and type of pesticide on the yield of oranges.
   
  

  Data File	SAS Source Code and Output	R Source Code and Output
  orange.txt	orange.sas SAS Output SAS Plots: a b c d1 d2 d3 e f1 f2 f3	orange.R R Output R Images: orange.pdf

  
  
• Orange Example -- Two-way ANOVA model with and without Interaction. Simplified version of Orange Example
   
  Investigate the effect of tree variety and type of pesticide on the yield of oranges.
   
  

  Data File	SAS Source Code and Output	R Source Code and Output
  orange.txt	orange2.sas SAS Output Interaction Plot1 Interaction Plot2	orange2.R R Output

  
  
• Efficiency Example -- Comparison of Mouse and Keyboard Efficiency
   
  Perform a mixed effects ANOVA analysis of the HCI dataset efficiency.txt. The dataset contains these variables:
  

  Variable	Meaning
  time	Access time in milliseconds
  item	Menu or toolbar item
  correct	Whether item was selected properly (good, bad)
  cond	Selection condition (mk=select menu with keyboard shortcut, mm=select from menu using mouse, tm=select from toolbar using mouse.)
  id	Participant ID
  trial	Trial number
  block	Block number

  
  Paper in the Proceedings of the Human Factors and Ergonomics Society:
  Comparison of Mouse and Keyboard Efficiency by Omanson, et. al.
  The paper uses a log transform on the dependent variable before performing the mixed effects ANOVA analysis.
   
  

  Data File	SAS Source Code and Output
  efficiency.txt	efficiency.sas SAS Output SAS Graph

Logistic Regression

• Discrim Example
   
  Model the probability of being rejected for a job application, given the independent variable settings.
   
  

  Data Set	SAS Source Code	R Source Code
  discrim.txt	discrim.sas SAS Output	discrim.R R Output

  
  
• Depress Example
   
  Model the probability of medical depression, given the independent variable settings. The estimation algorithm does not converge and produces 0 and 1 probabilities
   
  

  Data Set	SAS Source Code	R Source Code
  depress.txt	depress.sas SAS Output	depress.r R Output

  
   
• Remission Example
   
  Model the probability of complete cancer remission.  Use stepwise regression to select the variables: SLENTER=0.3, SLSTAY=0.35.
  Here is reference for more details:
  www.okstate.edu/sas/v7/sashtml/books/stat/chap35/sect43.htm
   
  

  Data Set	SAS Source Code	R Source Code
  remission.txt	remission.sas SAS Output	remission.r R Output

  
  

Goodness-of-fit Tests

• FairCoin1 Example
  Flip a coin 50 times recording 1 for head, 0 for tail.  Perform a χ² test for whether the probability of obtaining a head equals the probability of obtaining a tail.
   
  

  SAS Source Code
  fair-coin1.sas SAS Output

  
  
• FairCoin2 Example
  Perform the same χ² test as in the FairCoin1 Example, but analyzing the contingency table rather than the raw data.
   
  

  SAS Source Code	R Source Code
  fair-coin2.sas SAS Output	fair-coin2.r R Output

  
   
• FairCoin3 Example
  Perform a one-sample t-test for testing whether the proportion of heads is 0.5.
   
  

  SAS Source Code	R Source Code
  fair-coin3.sas SAS Output	fair-coin3.r R Output

• Cold1 Example
  Perform a chi-squared test for independence to see if a new drug makes a difference in the number of subjects contract a cold.
   
  

  SAS Source Code	R Source Code
  cold1.sas SAS Output	cold1.r R Output

  
  
• Cold2 Example
  Perform a t-test test for independence to see if a new drug makes a difference in the number of subjects contract a cold. Since n < 30 this test is very close to the χ² tests of the FairCoin1 and FairCoin2 Examples. It can be shown that the square of the z statistic equals the X² statistic of the χ² tests.
   
  

  SAS Source Code	R Source Code
  cold2.sas SAS Output	cold2.r R Output

  
  
• Cold3 Example
  Build a logistic regression model for predicting the probability of contracting a code based on whether the subject takes ascorbic acid or not.
   
  

  SAS Source Code
  cold3.sas SAS Output

• Condo Example
   
  Model the sale price of condo units at auction. See case study 1 of the 8/17 Notes and pages 447 to 465 of the Mendenhall and Sincich textbook for details.
   
  

  Data Set	SAS Source Code	R Source Code
  condo.txt	condo.sas SAS Output	condo.R R Output

  
  
• Trucking Example
   
  Model the log of the trucking cost using 7 models.  See the case study 2 of the 8/17 Notes and pages 345 to 354 of the Mendenhall and Sincich textbook for details.
   
  

  Data Sets
  trucking.txt (134 observations)
  trucking4.txt (448 observations)