While on Cape Cod this summer and when away from beach, I enjoyed some work-unrelated fun with Tableau. My beach reading included this article: http://www.theinformationlab.co.uk/2014/03/27/radar-charts-tableau-part-3/ by Andrew Ball and I decided to create my own Radar. When I show it to coworkers later, they suggested to me to publish it (at least the fun with Polygons, Path and Radars) on my blog. I may reuse this Radar chart for our internal Web Analytics.
Natural Order of Points and Segments in Line.
Many visualization tools will draw the line chart, its datapoints and connecting line segments between datapoints in natural progressing order – repainting them from left to right (horizontal ordering by Axis X) or from bottom to upside
(vertical ordering by Axis Y) or vice versa.
Path as the method to break the Natural Order.
Some demanding visualizations and users wish to break the natural repainting and drawing order and Tableau allows to do that by using the Path as the method to order the datapoints and line segments in Lines and Polygons. A Collection of increasing Ordering Numbers (Pathpoints) for each Datapoint in Line defined a Path for drawing and connecting datapoints and segments of that Line (or Polygon). Each Pathpoint can be predefined or calculated, depends on mplementation and business logic.
Changing the Natural Order can create “artificial” and unusual situations, when two or more datapoints occupying the same pixels on drawing surface but have very different Pathpoints (example can be a Polygon, when Line ends in the same point it starts) or when two or more Line Segments intersecting in the same Pixel on screen (example can be the Center of the letter X ).
Radar Chart has 2 parts: Radar Grid (background) and Radar Polygons (showing repetitive Data Patterns, if linear timeline can be collapsed into circular “timeline”). Radar Grid has Radials (with common Center) and Concentric Rings.
Polygons optionally can be filled with (transparent) color. For future Discussion let’s use the RMax as the maximal possible distance between the Center of Radar Grid (in case of Radar Grid) or the Center of Radar Polygon (in case of Radar
Polygon) and the most remote Datapoint shown in Radar Grid or Polygon respectively. We will use the “normalized” statistics of Visits to typical Website to visualize the hourly and daily (by day of the week) patterns of Web Visitations. By
normalization we mean the removal of insignificant deviations from “normal” hourly and daily amounts of Web Visits. For complete obfuscation we will assume for Demo purposes that RMax = 144.
Radar Radial Grid.
Radial Grid contains a few Radiuses (equidistant from each other) and we will draw each Radius as 3-point line where Starting and Ending points of each line are identical to each other and collocated with the Center of Radar. For Demo Web Visitation Radar we will use Radial Grid with 8 Radiuses, corresponding to the following hours of the complete 24-hours day: 0, 3, 6, 9, 12, 15, 18, 21:
For example see the Radius, corresponding to HOUR = 3 (below in light brown, other Radiuses greyed out on that image):
And for that Radius we are using (redundantly) the following 3 datapoints:
Concentric Rings for Radar Grid.
For Demo Radar we will use 4 Concentric Rings, corresponding to 25%, 50%, 75% and 100% levels of maximum visitation per hour:
Each ring is a line with 25 datapoints, where Starting and Ending Points collocated/equal. For example, dataset for external Ring (red line above) looks like this:
When Radials and Concentric Rings collocated and overlaid they represent the Radar Grid, ready to be a background for Radar Chart:
For Demo purposes we use only 2 Polygons – one (largest) representing average Hourly Visits during Weekday and 2nd Polygon representing average Hourly Visits during Weekend day. For Websites which I observed the minimum number of visits happened around 1 AM, so you will see both Polygons are slightly rotated clockwise and slightly shifted up from the Center of Radar Grid to reflect the fact that the minimum number of visitors (even around 1 AM) is slightly more then 0. Each Radar Polygon (in our Demo) has 25 Data Points with Starting and Ending Points collocated at 1AM. Here is a Weekday Polygon, overlaid with Radar Grid:
Here are the data for Weekday Polygon:
Here is a Polygon for Weekend day, overlaid with Radar Grid:
When Radar Grid and Radar Polygons overlaid (Polygons transparent but on top of
Grid) we will get the Radar Chart. Please note that Centers of Radar Grid and Radar
Polygons can have different locations:
I published Tableau workbook with this Demo Radar Chart and Radar Data here: