correlation analysis

[willaume@…]

I don't understand the use of the methods :

• .PCC
• .PRCC
• .SCC
• .SRCC

The first two methods give an error message which seems to be a bug. The second two methods give another error message. (see attached files)

[anonymous]

It follows from the definition of PCC and PRCC taht these quantities cannot be computed if the dimension of the input sample is not at least equal to 2, see OpenTURNS_ReferenceGuide, section C'. Nevertheless, OpenTURNS should complain about the wrong input sample dimension and it doesn't, which is a bug. It will be fixed in the next maintenance release and the next release.

For the two files SCC.py and SRCC.py, there is a mispelling between SCC (does not exist) and SRC (Standard Regression Coefficient), and between SRCC (does not exist) and SRRC (Standard Rank Regrassion Coefficient).

[willaume@…]

For the two files SCC.py and SRCC.py, there is a mispelling between SCC
(does not exist) and SRC (Standard Regression Coefficient), and between
SRCC (does not exist) and SRRC (Standard Rank Regrassion Coefficient).

So there is a bug in the TUI user manual.

[regis.lebrun@…]

Yes: there is a bug in the documentation. By the way, it is clearly stated in the exact same documentation that the dimension of the firt sample must be > 1 for PCC and PRCC ;-), and it is also (wrongly) stated that the same restriction occurs for SRC and SRRC. Nevertheless, I added a test to CorrelationAnalysis class and I requalify this ticket as a documentation issue.

[willaume@…]

In my documentation I have no sign ">", just a reverse "?".

[regis.lebrun@…]

Something bad during the LaTeX compilation? Can you give the pdflatex compilation log?

[conty]

Replying to regis.lebrun@eads.net:

Something bad during the LaTeX compilation? Can you give the pdflatex compilation log?

Users from Phimeca Engineering use Debian packages, they don't have any compilation logs. As maintener, I can deliver you the compilation log of building package process. Unfortunately I have removed all logs of 0.11.1 building process but the 0.11.2 is coming (very) soon and I will keep them.

[conty]

Ubuntu/Gutsy in amd64 compilation log files attached.

[conty]

Ubuntu Gutsy Gibbon amd64 compilation log file.

[anne.dutfoy@…]

I fixed the typo in the documentation concerning these methods.

[willaume@…]

In the TUI user manual I downloaded a few minutes ago :

• the sample 1 for PCC and PRCC methods is still specified of dimension greater or equal to 1.
• SRCC is not corrected

[regis.lebrun@…]

This time, it is ok. I also updated the online documentation.

Régis