By default, odfWeave renders data frames, matrices, and vectors in an attractive table format. The odfTable() function can be used to format tables with a higher degree of precision and control. The function produces XML code, so be sure to specify result=xml in code chunks employing this function. Unfortunately, the xtable() function doesn’t work with odfWeave.

Once you have your report in ODF format, you can continue to edit it, tighten up the formatting, and save the results to the ODT, HTML, DOC, or DOCX file format. To learn more, read the odfWeave manual and vignette. Additional information, including a tutorial on document formatting with odfWeave, can be found in the Examples folder installed in the odfWeave directory in your R library. (The function

.libPaths() will display your library location.)

22.5 Creating dynamic reports with R and Microsoft Word

For good or ill, Microsoft Word is the standard for report writing in corporate settings. You’ve already seen how to create a Word document from a Markdown file using rmarkdown (section 22.2). In this section, we’ll look at a method for creating dynamic reports by embedding R code directly into Word documents using the R2wd package. The methods in the section will only work on Windows platforms (sorry, Mac and Linux users).

To follow the examples in this section, you’ll need to install the R2wd package (install.packages("R2wd")). R2wd also requires the RDCOMClient package, available from the Omega Project for Statistical Computing.

At the time of this writing, RDCOMClient must be installed from source. First be sure Rtools is installed (http://cran.r-project.org/bin/windows/Rtools). Next, download the source file (RDCOMClient_0.93-0.tar.gz) from www.omegahat.org/RDCOMClient. Note that the version numbers are likely to change over time. Finally, install the package using

install.packages(RDCOMClient_0.93-0.tar.gz, repos = NULL, type = "source")

The R2wd package provides functions that allow you to create a blank Word document; insert sections and titles; insert text, tables, and graphs; add formatting; and save the results. Although the package is versatile, building and formatting Word documents programmatically can be time-consuming.

The easiest way to create a dynamic report in Word using the R2wd package is a two-step process:

1Create a Word document that contains bookmarks indicating where you want your R results to be placed.

2Create an R script that inserts the results in the Word document at the bookmarked locations and saves the finished document.

Let’s try this approach.

Open a new Word document, and call it salaryTemplate2.docx. Add the text and bookmarks displayed in figure 22.6. (In reality, the bookmarks in figure 22.6 aren’t

Listing 22.3 salary.R: R script for inserting results in salary.docx

First, the salary2Template.docx file is opened. If Word isn’t running, it’s automatically launched b. Next, the data analyses are performed. The cursor then is moved to the bookmark named n, and the sample size is inserted c.

Next, the cursor is moved to the bookmark named aovTable and the ANOVA results are inserted as a Word table d. Because R2wd doesn’t support the xtable() function, it’s important that the table be an R data frame, a matrix, or an array. Options can control the text and location of the table caption, font size, and table formatting. Try autoformat = 1, 2, 3, and so on, to see the various formats available. Currently there is no way to suppress the caption.

Two of the statements in the ANOVA code require additional explanation. The aovTable object is a data frame containing the two-way ANOVA results. The round() function is used to limit the number of decimal places printed in the table. The statement

aovTable[is.na(aovTable)] <- ""

is a trick that replaces NAs with blanks. This is necessary because there are no values in the F and Pr(>F) columns for the Residuals row, and you don’t want NA to print in these cells of the table.

The cursor next moves to the bookmark named effectsPlot. The wdPlot() function requires that the user specify a plotting function. Here, the myplot() function produces an effects plot via the allEffects() function in the effects package e.

The wdPlot() function supports method="bitmap" and method="metafile". Use metafile whenever possible—it looks better in a Word document. Unfortunately, the metafile option doesn’t support transparency, so you have to use the bitmap option when transparency is present. You’re most likely to encounter transparency when using the ggplot2 package to produce graphs.

When the code in salary.R is executed, it runs the R code, inserts the results into salaryTemplate2.docx, and saves the finished Word document as salaryReport2.docx f. Then the Microsoft Word application quits. The resulting document is displayed in figure 22.7.

Sample Report

Introduction

A two-way analysis of variance was employed to investigate the relationship between gender, academic rank, and annual salary in dollars. Data were collected from 397 professors in 2008. The ANOVA table is given in Table 1.

Table 1 Two-way Analysis of Variance

The interaction between gender and rank is plotted in Figure 1.

Figure 1 Mean Effects Plot

Figure 22.7 Final report in DOCX format (salaryReport2.docx)