Results of the Readers’ Survey

 First of all, let me say “Thank You” to all of the 357 people who completed the survey. I was hoping for 100, so needless to say the response blew away my expectations. This endeavor seems like a worthwhile effort to do once a year. Next year I will refine the questionnaire based on what I’ve learned. Perhaps this can this can evolve into more of a report on the state of the R community and less on me trying to determine what should be the focus of my blog.

• Age
• Gender
• Education Level
• Employment Status
• Country
• Years of Experience Using R
• Ranking of Interest to See Blog Posts on Various Analysis Techniques (7-point Likert Scale)
• Feature Selection
• Text Mining
• Data Visualization
• Time Series
• Generalized Additive Models
• Interactive Maps
• Favorite Topic (select from a list or write in your preference)

gender
Female   Male 
    28    329
> barchart(education, main="Education", xlab="Count") #barchart of education level

This basic barchart in lattice needs a couple of improvements, which is simple enough.  I want to change the  color of the bars to black and to sort the bars from high to low.  You can use the table and sort functions to    accomplish it.
> barchart(sort(table(education)), main="Education", xlab="Count", col="black")

There are a number of ways to examine a Likert  Scale.  It is quite common in market research to treat a Likert   Scale as continuous data instead of ordinal.  Another technique is to report the “top-two box”, coding the      data as a binary variable.  Below, I just look at the frequency bars of each of the questions in one plot using   ggplot2 and a function I copied from the excellent ebook by Winston Chang, 
Cookbook for R, http://www.cookbook-r.com/ .
You can easily build a multiplot, first creating the individual plots:
> library(ggplot2)

> p1 = ggplot(survey, aes(x=feature.select)) + geom_bar(binwidth=0.5) + ggtitle("Feature Selection")               + theme(axis.title.x = element_blank()) + ylim(0,225)
> p2 = ggplot(survey, aes(x=text.mine)) + geom_bar(binwidth=0.5) + ggtitle("Text Mining")                                + theme(axis.title.x = element_blank()) + ylim(0,225)
> p3 = ggplot(survey, aes(x=data.vis)) + geom_bar(binwidth=0.5, fill="green") + ggtitle("Data Visualization)    + theme(axis.title.x = element_blank()) + ylim(0,225)
> p4 = ggplot(survey, aes(x=tseries)) + geom_bar(binwidth=0.5) + ggtitle("Time Series")                                     + theme(axis.title.x = element_blank()) + ylim(0,225)
> p5 = ggplot(survey, aes(x=gam)) + geom_bar(binwidth=0.5) + ggtitle("Generalized Additive Models")           + theme(axis.title.x = element_blank()) + ylim(0,225)
> p6 = ggplot(survey, aes(x=imaps)) + geom_bar(binwidth=0.5) + ggtitle("Interactive Maps")                             + theme(axis.title.x = element_blank()) + ylim(0,225)
Now, here is the function to create a multiplot:
> multiplot <- function(..., plotlist=NULL, file, cols=1, layout=NULL) {
+ require(grid)
+ # Make a list from the ... arguments and plotlist
+ plots <- c(list(...), plotlist)
+ numPlots = length(plots)
+ # If layout is NULL, then use 'cols' to determine layout
+ if (is.null(layout)) {
+ # Make the panel
+ # ncol: Number of columns of plots
+ # nrow: Number of rows needed, calculated from # of cols
+ layout <- matrix(seq(1, cols * ceiling(numPlots/cols)),
+ ncol = cols, nrow = ceiling(numPlots/cols))
+ }
+ if (numPlots==1) {
+ print(plots[[1]])
+ } else {
+ # Set up the page
+ grid.newpage()
+ pushViewport(viewport(layout = grid.layout(nrow(layout), ncol(layout))))
+ # Make each plot, in the correct location
+ for (i in 1:numPlots) {
+ # Get the i,j matrix positions of the regions that contain this subplot
+ matchidx <- as.data.frame(which(layout == i, arr.ind = TRUE))
+ print(plots[[i]], vp = viewport(layout.pos.row = matchidx$row,
+ layout.pos.col = matchidx$col))
+ }
+ }
+ }
And now, we can compare the 6 variables on one plot with one line of code with the clear winner data              visualization.
> multiplot(p1, p2, p3, p4, p5, p6, cols=3)

The last thing I will cover in this post is how to do maps of the data on a global scale and compare two  packages: rworldmap and googleVis. > library(rworldmap) > #prepare the country data  > df = data.frame(table(country)) > #join data to map > join = joinCountryData2Map(df, joinCode=”NAME”, nameJoinColumn=”country”, suggestForFailedCodes=TRUE) #one country observation was missing > par(mai=c(0,0,0.2,0),xaxs=”i”,yaxs=”i”) #enables a proper plot This code takes the log of the observations. Without the log values a map was produced that was of no  value. It also adjusts the size of the Legend as the default seems too large in comparison with the map.  > map = mapCountryData( join, nameColumnToPlot=”Freq”, catMethod=”logFixedWidth”, addLegend=FALSE) > do.call( addMapLegend, c(map, legendWidth=0.5, legendMar = 2)) Now, let’s try googleVis. This will produce the plot in html and it is also interactive. That is, you can hover your mouse over a country and get the actual counts. > library(googleVis) > G1 <- gvisGeoMap(df, locationvar=’country’, numvar=’Freq’,  options=list(dataMode=”regions”)) > plot(G1)

That's it for now.  I absolutely love working with data visualization on maps and will continue to try and            incorporate them in future analysis.  I can't wait to brutally overindulge and sink my teeth into the ggmap       package.  Until next time, Mahalo.