This week, the weather in Madison has miserably hot, and thankfully the campus air conditioning seems to be back online.

So, is it really that much worse than it was a week ago?

I wanted to look at the weather data, and was delighted to find out how much of the work had already been done for me through a variety of R functions and packages. I found a very nice function to download  data from weatherunderground.com, which in turn uses the XML package to scrape tables of data off of web pages.  I downloaded hourly weather reports over the past 50 days from the weather station atop the UW AOSS building. I used plyr to take the maximum and minimum daily values for humidity, temperature, dew point, wind speed, etc.

I thought Chernoff faces could qualitatively convey weather patterns over time, and I had two options in R.  Below are the last 50 days of Madison weather (with the date and high temp (F) as labels).  Bigger, redder faces show hotter days, and high humidity is represented by a frown. Bigger hair means a higher dew point, and up-swept hair indicates a relatively windy day.

Click for larger

I think the figure does a good job showing that the last 4 days were hot, muggy, and miserable; however, I am not totally sold on Chernoff faces. Facial features are not always intuitively linked to the features of  being presented, like precipitation being represented by eye size. (Tears? Seems a bit forced.)  I also worry that certain features are more noticeable than others, like the face color–which is probably fine in this case.