Jump to content

Scatter plot - highlight bad time series data based on associated QC column.


Sean Flaherty 2

Recommended Posts

Hello,

We have a situation where we are trying to identifyindividual data points on a scatter plot based upon values in a secondary column.We would like toplot multiple columns of time series data each with an associated quality control column for each time stamp.

In our example, we are using a scatter plot with2time series data columns (Col_1, Col_2) each with a seperate, associated column containinga quality control flag (Col_1_QC, Col_2_QC). (See attached image_1.)

We would like to be able to plot the 2data columns on the Y axis and change the color and/or shape of an individualline's data point based upon the value their associatedquality control column. We have been able to color all data column points based upon a single quality control column but can't figure a way to "pair" these columns.

We also triedcreating a calculated column for each data column of just thebad values based on the QC columnsand plot that data series on top the existing graph using a different color. However, we use the line connection to show order of the time seriesdata on the graphwhichalso gives us lines connecting the error values (see red lilne in attached graph_errorlines image). We can't figure out how to shut the line connection down for just those columns.

I attached an image showing an example of what we are trying to do with the red "X" making values where thevalue in the associated QC columnfor that timestamp was greater than 1.

Any suggestions or ideas would be appreciated.

Thankyou and

Link to comment
Share on other sites

Thank you, Gaia.

For our scatter plot, we ended up using the "Line from Column Value" feature under "Lines and Curves" instead of"Line Connection". This allowed us to create linesfor the GHI_1 and GHI_2 columns that tracked the scatter plot pointsand also excludethe GHI_1_Error and GHI_2_Error columns so those appeared only as discreet points.

Hopefully thissolution helps someone else.

Kind regards,

Sean

Link to comment
Share on other sites

Hello Gaia,

 

Thanks for the response.  I have attached an example.dxp of what we are trying to accomplish. 

 

We would like to have a single graph that contains both lines in the "Individual Graphs" tab.  When I create a single graph as in the "CombinedWithErrors" tab, it is close but we need to figure out a way to get rid on the red lines between the  error values.  

 

Thank you and kind regards,

 

Sean

Link to comment
Share on other sites

I cannot find a way to tag each column with its own error mark. Also if the markers for GHI_1 and GHI_2 were in different timestamps (not in your example) it would become impossible to tell which column is marked as error.

I had an alternative idea.

1-use a Combination Chart, add 4 Y columns: GHI_1, GHI_2, GHI_1_Error and GHI_2_Error

2-display GHI_1 and GHI_2 as lines

3-display GHI_1_Error and GHI_2_Error as bar chart (side-by-side)

The series is displayed as Avg(..) but that should be ok as you have one point per timestamp.

In your example the errors overlap, so the bars would hide one another in case both curves have an error, but you could mark the bar to show a details table

See screenshot.

Link to comment
Share on other sites

×
×
  • Create New...