Errata and Software Updates
A Windows Installer for R v 3.01
Newer versions of R and Rcommander do not always match the textbook. Here's a full Windows installer for the version used in the text, and that includes the BCA Plugin needed for the text. This installer replaces the complex installation procedure in section 3.2 of the text. The defaults work fine. You do have one customization option – whether you prefer one big window or separate windows for the various functions. The separate windows are usually easier to manage.
Unfortunately, there is no equivalent installer for Mac OS X.
If you do use the installation procedures in the text and get newer versions of R, here are a few differences:
- Tabbed dialogs
Several of the more complex dialog windows shown in the screenshots in this text have now been split into tabbed dialogs, with one tab typically labled "Options".
- Page 45-46
After entering the source command, a window will pop up: "Would you like to use a personal library instead?" Select Yes. This is followed by a second window that suggests a default directory. Select "Yes" for this directory as well. This is followed by the list of CRAN mirrors, as specified in the text and shown in Figure 3.14.
- Page 47,
After entering library(RcmdrPlugin.BCA), a window will appear requesting the installation of more packages, and a source to install packages from. Accept the default CRAN location and click OK. Additional packages will install. At that point the R Commander GUI will appear, disappear and reappear.
- Page 47
The RCommander window now appears slightly different from the text. Notably the script window now contains a tab "R Markdown" which is an authoring format to create documents and presentations from R. We will not use this feature.
Newer versions of Excel will have different interfaces from that shown here.
Page 75, 4th line
The sentence: "Among the 'Hyptoheses Tests' options, select only the chi-square test." is redundant, as this has been noted a few lines previously.
Page 78: Remove Records dialog is inefficient.
Records must be removed to calculate a chi-square statistic in the contingency table. The following link provides a description of the problem and short script to much more quickly remove records, and introduces the very useful subset command. Remove Records
Page 153, Table 6.2
The ninth entry in the first column (Monthly Giver) should be "Yes" and in the third column (Weight) should be "1".
Page 228, 3rd line
The instruction Summary == "Validation" should be Sample == "Validation".
If the centroids in Figure 10.10 do not appear, don't panic. Continue on to step 7, which shows you how to add the new cluster label variable, Wards4, to the data.
Once that is done, you can generate the cluster centroids-- which are just the means of the variables in each cluster-- with Explore and Test=> Summarize=> Numerical Summaries. Select all of the variables, check only the mean, click the Summarize by Groups button at the bottom, select Wards 4, and click OK twice.