Ensure the car and ggvis packages are loaded. Revisit the Salaries data frame you looked at in Exercise 24.1; inspect the help file ?Salaries to remind yourself of the present variables.
a. Produce an interactive scatterplot of salary on the vertical axis and the years of service on the horizontal axis. Employ radio buttons to color points according to either academic rank, research discipline, or sex. Use pipes to add_legend and add_axis to omit a legend title and to tidy up the axis titles, respectively.
b. A pipe to layer_densities (which you’ve not yet met) is used to produce kernel density estimates, similar to those appearing in Figure 24-5.
i. Use ggvis to create a static plot of kernel density estimates of salary distributions, split up according to academic rank. To do this, assign the salary variable to x and the rank variable to fill, followed by a pipe to group_by to explicitly instruct grouping by the rank variable. Lastly, piping to layer_densities (just use all default argument values in this instance) will generate the graphic. Your result should resemble the gg4 object from Exercise 24.1.
ii. Just like the width argument to layer_histograms is used to control the appearance of a histogram, the adjust argument in layer_densities is used to control the degree of smoothness of the kernel estimates. Reproduce the rank-specific kernel estimates from the previous plot, but this time, the graphic should be interactive—implement a slider button with a range of 0.2 to 2 and a label of "Smoothness" to control the smoothing adjustment. At your discretion, either suppress or clarify the axis and legend titles of the result.
Ensure you have the MASS package loaded, once more gaining access to the UScereal data frame. If you haven’t already done so, inspect the help file ?UScereal and re-create the cereal object exactly as specified in Exercise 24.1 (a). Then do the following:
c. Set up an object for radio buttons to choose among the manufacturer, the shelf, and the vitamins variables. Make sure the labels for each radio button are clear, and set up an appropriate title label for what will form the collection of options to color the points. Name the object filler.
d. Borrowing the sizer and opacityer objects created in Section 24.4 and using the object you just created in (c) to control fill, create an interactive scatterplot of calories on protein. Tidy up the axis titles and suppress the legend title for the point color fill. The result should essentially be the same, in terms of functionality, as the graphic appearing as the topmost screenshot in Figure 24-8.
e. Create a new object for the same radio buttons as specified in (c) that will control the shape of the points (in other words, the characters used to plot points). Modify the title label accordingly. Name this object shaper.
f. Finally, re-create the interactive scatterplot of calories on protein exactly as in (d), but this time additionally assigning shaper from (e) to the shape modifier in your call to ggvis. To prevent the legends for the two sets of radio buttons from overlapping each other, you need to add the following pipes to your code:
The first simply moves the legend for the shape modifier vertically downward, and the second eliminates the slight “animation delay” that occurs by default when switching between options in the interactive graphic. Once more, use additional calls to add_axis and add_legend to clarify or suppress axis and legend titles.