Descriptive Statistics needs cleanup (#18795) · Issues · ParaView / ParaView

Descriptive Statistics needs cleanup

Please clean up the Descriptive Statistics filter.

Example run:

5.6.0, Linux, remote server (NOT builtin server).
Load g1s1-10fps.e.16.[0-15]. All variables on. Apply.
Filters/ Descriptive Statistics.

So, what I expect to see is statistics (mean, median, max, min, possibly standard deviation) for each variable. So, here are modifications I want to see:

We want to see statistics over the whole DATASET, NOT each core or processor. (OK, if necessary, statistics per block is acceptable.) Users want to know about their dataset, not how the cluster partitioned their data!
Please add in, and optionally calculate magnitudes of vectors, default on. Optionally present the coordinates of vectors (such as Displacement), default off.
Please make M2, M3 and M4 an advanced output. In other words, standard deviation, skewness and kertosus should be advanced outputs.
Add a one sentence explination for each of M2, M3 and M4. I believe M2 is standard deviation, M3 is skewness and M4 is kertosus. https://www.paraview.org/Wiki/Statistical_analysis#Descriptive_Statistics.
Reorder the columns to be Block number, Row number, Variable name, Max, Min, Mean, Median, Standard Deviation, etc.
This filter needs to work correctly in parallel, taking into account processor boundaries. We don't want to double count the same points.

Don't forget that this filter will be used by statiticians that understand this stuff, but also by engineers, scientists and students that may not know this stuff. Remember a person that took statistics 10 years ago may know what they are looking for, but may not remember the terminology. Further, M2, M3 and M4 are generic enough that they are impossible to google.

Please clean up the Descriptive Statistics filter.

Example run:
* 5.6.0, Linux, remote server (NOT builtin server).
* Load g1s1-10fps.e.16.[0-15].  All variables on.  Apply.
* Filters/ Descriptive Statistics.

So, what I expect to see is statistics (mean, median, max, min, possibly standard deviation) for each variable.  So, here are modifications I want to see:

* We want to see statistics over the whole DATASET, NOT each core or processor.  (OK, if necessary, statistics per block is acceptable.)  Users want to know about their dataset, not how the cluster partitioned their data!
 
* Please add in, and optionally calculate magnitudes of vectors, default on.  Optionally present the coordinates of vectors (such as Displacement), default off.

* Please make M2, M3 and M4 an advanced output.  In other words, standard deviation, skewness and kertosus should be advanced outputs.

* Add a one sentence explination for each of M2, M3 and M4.  I believe M2 is standard deviation, M3 is skewness and M4 is kertosus.  https://www.paraview.org/Wiki/Statistical_analysis#Descriptive_Statistics.

* Reorder the columns to be Block number, Row number, Variable name, Max, Min, Mean, Median, Standard Deviation, etc.

* This filter needs to work correctly in parallel, taking into account processor boundaries. We don't want to double count the same points.

Don't forget that this filter will be used by statiticians that understand this stuff, but also by engineers, scientists and students that may not know this stuff.  Remember a person that took statistics 10 years ago may know what they are looking for, but may not remember the terminology.  Further, M2, M3 and M4 are generic enough that they are impossible to google.

Edited Feb 16, 2019 by W. Alan Scott

Discussion
Designs

The one place for your designs

To enable design management, you'll need to meet the requirements. If you need help, reach out to our support team for assistance.