Using a Scatter Plot as a Quality Control Tool

You can use Windographer's Flag By Scatter Plot window or Scatter Plot window, to help detect and track down many types of problems in your dataset. Using the ability to filter data can also be helpful to this process. To demonstrate, we will take a look at an scatter plot that shows data compromised by icing.

The scatter plot below shows the wind speed at 45m above ground versus the wind speed at 10m above ground in December. Normally the 45m wind speed should almost always exceed the 10m wind speed, but this scatter plot shows that the opposite occurred in many time steps. (These points appear highlighted in red.) Another thing that does not look right is that in several time steps one anemometer is recording near-zero wind speeds while the other is recording moderate wind speeds, between 3 m/s and 12 m/s. (These points appear highlighted in yellow.) Both of these features of the scatter plot hint at problems in the dataset.

We can use the filtering capability on the Scatter Plot window or Flag By Scatter Plot window to investigate further. For example, if you suspected that icing events caused the errant data points in the above scatter plot, you might want to filter those data according to temperature. To display data points only for times when the temperature was above freezing, you could use the following filter settings:

When we apply that filter to show only above-freezing data, all of the extraordinary data points disappear from the scatter plot, as shown below. This suggests that icing has indeed caused the apparent problems in the December data.

See also

Flag By Scatter Plot window

Scatter Plot window


Written by: Tom Lambert
Contact: windographer.support@ul.com
Last modified: August 15, 2012