mages' blog

The guys at RStudio have done a fantastic job with shiny. It is really easy to build web apps with R using shiny. With the help of Joe Cheng from RStudio we figured out a way to make googleVis work on shiny as well. This allows you to make use of the Google Charts Tools in your shiny app directly from R. What I present here are three initial examples which seem to work in most browsers. The third example even uses a neat trick to create an animated geo chart.



However, before we upload the next version of googleVis to CRAN we decided to present a preview of version 0.4.0 here, asking for feedback. It would not be fair on the guys behind CRAN to release something into the wild, only to be told by users within a few days that we missed something. Hence, you can get the new version of googleVis only from the download page of our project site for the time being.

You may have read the post on RStudio's blog that the shiny API changed slightly: the reactivePlot and reactiveText functions have been renamed to renderPlot and renderText with simplified input parameters. Thanks to Joe, there is now also a renderGvis function as part of the googleVis package, which works in very much the same way as the other two.

To run the following examples you need shiny version 0.4.0 and googleVis version 0.4.0 or higher.

Michael Bach, who is a professor and vision scientist at the University of Freiburg, maintains a fascinating site about visual illusions. One visual illusion really surprised me: the sigma motion.

The sigma motion displays a flickering figure of black and white columns. Actually it is just a chart, as displayed below, with the columns changing backwards and forwards from black to white at a rate of about 30 transitions per second.

The data of the World Bank is absolutely amazing. I had said this before, but their updated iPhone App gives me a reason to return to this topic. Version 3 of the DataFinder App allows you to visualise the data on your phone, including motion maps, see the screen shot below.

Screen shot of DataFinder 3.0

I was intrigued by the by the changes in life expectancy over time around the world. The average life expectancy of a new born baby in 1960 was only 52.6 years, and don't forget we are in 2012, 52 years later. Babies born in 2009 can expect to live to the age 69.4 years, an increase by nearly 17 years or 32%. That is remarkable. But these are average figures and vary a lot by country.

The world bank's online version of the map unfortunately lacks the the animation of the smartphone app. Thus I was keen to find out if I could reproduce something similar in R based on code and ideas provided to me by Manoj Ananthapadmanabhan and Anand Ramalingam last year. Those ideas had helped me to create the animated geo maps demo "AnimatedGeoMap" of the googleVis R package.

Fitting distribution with R is something I have to do once in a while, but where do I start?

A good starting point to learn more about distribution fitting with R is Vito Ricci's tutorial on CRAN. I also find the vignettes of the actuar and fitdistrplus package a good read. I haven't looked into the recently published Handbook of fitting statistical distributions with R, by Z. Karian and E.J. Dudewicz, but it might be worthwhile in certain cases, see Xi'An's review. A more comprehensive overview of the various R packages is given by the CRAN Task View: Probability Distributions, maintained by Christophe Dutang.

How do I decide which distribution might be a good starting point?

I came across the paper Probabilistic approaches to risk by Aswath Damodaran. In Appendix 6.1 Aswath discusses the key characteristics of the most common distributions and in Figure 6A.15 he provides a decision tree diagram for choosing a distribution:

Figure 6A.15 from Probabilistic approaches to risk by Aswath Damodaran

JD Long points to the Clickable diagram of distribution relationships by John Cook in his blog entry about Fitting distribution X to data from distribution Y . With those two charts I find it not too difficult anymore to find a reasonable starting point.

Once I have decided which distribution might be a good fit I start usually with the fitdistr function of the MASS package. However, since I discovered the fitdistrplus package I have become very fond of the fitdist function, as it comes with a wonderful plot method. It plots an empirical histogram with a theoretical density curve, a QQ and PP-plot and the empirical cumulative distribution with the theoretical distribution. Further, the package provides also goodness of fit tests via gofstat.

Suppose I have only 50 data points, of which I believe that they follow a log-normal distribution. How much variance can I expect? Well, let's experiment. I draw 50 random numbers from a log-normal distribution, fit the distribution to the sample data and repeat the exercise 50 times and plot the results using the plot function of the fitdistrplus package.


I notice quite a big variance in the results. For some samples other distributions, e.g. logistic, could provide a better fit. You might argue that 50 data points is not a lot of data, but in real life it often is, and hence this little example already shows me that fitting a distribution to data is not just about applying an algorithm, but requires a sound understanding of the process which generated the data as well.