I have a dataset including wind speed and direction. I would like to show the average wind speed in different wind direction with equal intervals (see attached). I have already binned the wind direction data into equal intervals. I also converted the wind speed and wind direction to components 'u' and 'v' using u = ws * cos(θ) and v = ws * sin(θ). I only need help in making a plot similar to the attached.
Accepted Solutions
This combines three of my least favorite things in dataviz (circular encoding, rainbow coloring, and means out of context), so of course I have to try it out!
The simplest thing to do is probably to use the computed x/y points and make a scatter plot of colored dots around a circle.
A fancy solution is to make a custom shape file where each shape is a wedge of the circle. Sometimes people do that for wafer maps so that each component can have a color.
For the circle, you have to know how many wedges you want in advance and refer to each by name. Here's what that can look like (I couldn't bring myself to use the rainbow):
Attached is a script that makes three data tables for this chart. Two data tables comprise the shape descriptions and the third one has the data, most random in this case. As long as you're happy with the wedges, you can use those files for different data tables. The trick is to set a column property called Shape Use in your data table to point to the shape tables.
Having gone wild and making a polar chart, I must at least point out this paper, Graphical Inference for Infovis, by Hadley Wickham et al, which has an incidental result where Cartesian coordinates were more effect than polar coordinates for wind data. Admittedly, they were unwilling to simplify down to a mean and were trying to show all the data.
I thing I can get closer than I demonstrated to the Wind chart you want. All I have done is to define 17 categories, and use the Value Colors column property to set the colors.
I didn't work too hard on getting the blues as dark as your chart, but all of those colors can be referenced. Attached is the data table with the Value Colors set, and below is the script I used to produce the chart.
Graph Builder(
Variables( X( :slice ), Y( :value ) ),
Elements( Pie( X, Y, Legend( 3 ), Pie Style( "Ring" ) ) )
);This combines three of my least favorite things in dataviz (circular encoding, rainbow coloring, and means out of context), so of course I have to try it out!
The simplest thing to do is probably to use the computed x/y points and make a scatter plot of colored dots around a circle.
A fancy solution is to make a custom shape file where each shape is a wedge of the circle. Sometimes people do that for wafer maps so that each component can have a color.
For the circle, you have to know how many wedges you want in advance and refer to each by name. Here's what that can look like (I couldn't bring myself to use the rainbow):
Attached is a script that makes three data tables for this chart. Two data tables comprise the shape descriptions and the third one has the data, most random in this case. As long as you're happy with the wedges, you can use those files for different data tables. The trick is to set a column property called Shape Use in your data table to point to the shape tables.
Having gone wild and making a polar chart, I must at least point out this paper, Graphical Inference for Infovis, by Hadley Wickham et al, which has an incidental result where Cartesian coordinates were more effect than polar coordinates for wind data. Admittedly, they were unwilling to simplify down to a mean and were trying to show all the data.
OK, so @XanGregg solution may be the best.......but there is a Pandemic going on, and I have too much time on my hands. So I have revisited my initial proposed solution, and formalized it into a script that takes the 16 directional wind speeds, and takes it all the way though the final graph. It uses the 0 to 450 color gradient, and has the same legend as the sample graph. Yup, this is my silliness, but it does work using Graph Builder's pie chart, plus some enhancements. Attached is the data table and the script.
