Re: cross tables in reports?

[Stanley A. Klein]

Derek -

I did some web searching on the relevant topics.  I also have from time to
time over the years built regression-based models and done statistical
analyses.  In my experience regression-based modeling is as much art as
science.  

It appears that OLAP is one of a number of technologies that people use for
strategic business analysis.  Traditional statistical packages are also
used. SAS and R are examples of these traditional packages.  Both OLAP and
traditional statistics require the analyst to identify the processing to be
used on the data, at least by selecting the particular analysis and the
relevant subsets of the data.  There are also AI-based approaches under
which the analytical tool searches for patterns in the data "on its own."  

I detected a lot of "buzzwordism" in what I saw.  I'm sure every one of
these approaches has its pros and cons.  I'm also sure that when this kind
of stuff is demo'ed it looks very "gee whiz" especially when stuff like 3D
graphics are included.

I think there are really two issues here.  One issue is whether a
free/open-source OLAP package can be built using A+.  I think that issue
stands on its own.  The people who can do that kind of thing would need to
be comfortable with both the Iverson notation (that is the basis of APL)
and the OLAP-type algorithms.  APL is a tricky language to learn.  But
aren't they all.  :)  It would also help if appropriate prior work were
available in the relevant APL and A+ repositories.  Perhaps it is.

The other issue is which of these approaches GNUe can support.  I hope the
answer is all of the above.  We can clearly support an interface to R,
which provides the traditional statistical approach.  

This thread stated with a question about crosstabs, which I think of as a
technique related to areas such as survey research and what are known in
statistics as "contingency tables."  That's why I suggested R.  "Pivot
tables" seems to be nothing more than M$'s buzzword for the same kind of
thing.

It is amazing in this business how much "new" technology is just a version
of the old technology with a name that nobody ever heard before.  It
happens all the time and has been going on for years.  Not exactly snake
oil or ginsu knives, but I think some aspects of the sales approach are
familiar.  :-)  


Stan Klein


At 03:52 PM 11/2/2002 -0700, Derek Neighbors wrote:
>On Sat, 2002-11-02 at 03:47, Stanley A. Klein wrote:
>> BTW, there appears to be work ongoing on GUI interfaces for R, but it
>> doesn't seem to be very far along.  It might be nice to let the R folks
>> know that GNUe might be a good candidate for that.
>
>I still think A+ is the best for this.  A+ team could create an generic
>OLAP framework, much like Hyperion, Adaytum and the likes.  Then it
>would just be another tool in the tool box.
>
>To do general crosstab type of reports I dont think an additional
>package is necessary.  The kind of financial 'cube' data that is
>discussed that falls into realm of data warehousing, I think we need to
>come at from an OLAP front.  If someone were to write an OLAP engine in
>R great.  I think it is more likely that someone will do so in A+.
>
>In fact, recently someone actually posted code to the A+ list that was a
>simple graphically OLAP package.  I use an OLAP package every day at
>work that is written in a APL Dialect called DyAlog.  
>
>-- 
>Derek Neighbors
>GNU Enterprise
>http://www.gnuenterprise.org
>address@hidden
>
>Was I helpful?  Let others know:
> http://svcs.affero.net/rm.php?r=dneighbo
>
>Attachment Converted: "c:\internet\eudora\attach\signatu1.asc"
>