CanCor (Canonical Correlations)

[EViews Gareth]

This thread is about the CanCor add-in that calculates canonical correlations between two group objects.

[JHaltmaier]

When I try to run CanCor I get an error message saying that !COLS is not defined or is an illegal command.

[EViews Gareth]

Could you provide the workfile you're using?

[JHaltmaier]

I can't because it contains confidential data. However, I have tried to use cancor in several different workfiles and with different groups and I always get the response that !COLS is not defined or is an illegal command, which seems to be a programming problem. Is that in fact an EVIEWS command? I can't find it in the command reference. Do you have workfiles in which cancor works? Maybe you could send one to me. Thanks.

[EViews Gareth]

If I run the following program to generate a workfile, then run the CanCor add-in, using GY and GX as the two groups, it works fine:

Code: Select all

rndseed 1 create u 100 series x1=nrnd series x2=nrnd series y1=nrnd series y2=nrnd group gx x1 x2 group gy y1 y2

[JHaltmaier]

I copied and pasted your program, ran it to create a workfile, tried to use CanCor and got exactly the same error message: !COLS is not defined or is an illegal command.

[JHaltmaier]

I also just discovered that it is creating tables called cancorr, cancorr01, etc. (apparently a new one every time I try to run it) that are empty.

[EViews Gareth]

What's the build date of your copy of EViews (help->about eviews)?

[JHaltmaier]

April 20 2010

[EViews Gareth]

Ouch, nearly a year out of date. You should update it (help->EViews update).