Re: WHY

hrlngrv_at_aol.com
Date: 12/28/04 

Date: 27 Dec 2004 16:11:02 -0800

aaron_kempf@hotmail.com wrote...
...
>RE: "No matrix arithmetic functions. "
>
>MDX stands for multiple-dimensional expressions.. if you're really
trying to
>claim that i can't build a 2-dimensional matrix and do whatever i
want-- I
>can build matrixs with 127 different dimensions.. i can build a
matrix--
>that instead of just going up/down and left/right-- i can go in 127
>different directions

Stick with 2D matrices for now. How would you invert one? How would you
multiply two conformant ones? How would you calculate the determinant
of one? Those Excel functions aren't supported. Neither are LINEST or
LOGEST.

As for higher dimensions, I'd prefer to write my tensors in Mathematica
or MatLab (or SciLab) than deal with individually indexed entries in
some hypergrid.

>Can you build a 3-dimensional matrix in Excel?? A pivot
>HAHAHAHAHAHAHAHAHAHAHAHAHAHAH get real kid

I don't use 3D (or higher D) arrays much in Excel. I do use
multidimensional data in Mathematica and APL, which are the more
appropriate tools for what I do. (Heck, I never made much if any use of
3D in 123 beyond relative work*** addressing.)

>you can't do a damn thing with more than 3 or 4 dimensions... i throw
around
>billions of records in 20 directions before breakfast every day.
...

ed speak: s/around/up/

Who needs to? Multiple dimensions aren't often needed or even useful,
at least not in what I do, and when they are, there are functional
relationships between the dimensions that make certain transformations
that reduce the number of dimensions optimal.

If you're summing and averaging sales by product by distribution
channel by region by quarter, then multiple dimensions (and caching)
make sense for something that simple. If you're doing any sort of trend
analysis, e.g., inventory control, you'd be pulling off slices in 1D
and 2D for analysis, at which point OLAP's benefits become secondary to
MDX's lack of built-in stats functions.

>You can do any of that stuff that you can in excel thru either
>
>VBA
>SQL Server
>or
>MDX
>
>Anyone that tells you otherwise is trying to sell you anything
...

How would you generate 1,000,000 random deviates from a lognormal
distribution with parameters mu = 8.5 and sigma = 2.0 using MDX? Yes,
this can be done in Excel, and no, it doesn't need to be stored in a
single column.

>And here is a listing from SQL Server Books Online:
...

OK, it does seem to include most of them, though not the ones
previously mentioned. Odd that MDX includes Median as one of its own
functions but not Mode, since Mode is pretty darn simple in SQL.

Select Max(x)
>>From (Select Count(y) As x From Tbl Group By y);

So, can you pass any old OLAP object to any Excel function, or do you
have to extract individual numbers from OLAP objects before you can use
Excel functions? If the latter, that interface would necessarily slow
things down considerably.

>And it's funny.. i looked up about linest.. I think that it is
hilarious
>that you would WANT to get an array from a function.. The bottom line
is
>that sub-queries are a LOT more powerful than arrays..

Getting arrays (or lists) from functions is the core of most functional
programming languages, like APL, Lisp, Scheme. Excel was the first
spread*** to adopt this concept. I can see why you find it funny -
you don't understand its use. And it beats 123's @REGRESSION, which
needs convoluted 3rd argument codes to return individual pieces of what
LINEST returns.

If you need a subquery, they may be better than arrays. If you need the
matrix product of two conformant 2D arrays, subqueries just don't cut
it. You don't understand the problem domain (again).

>If you weren't SCARED TO BE SEEN AS A DATABASE DEVELOPER you would
know
>that..

If you weren't so confused into believing that databases are the only
tool you'll ever need, and so ignorant about what OLAP/MDX can't do,
you might have a chance to learn much of what you don't know.

>And don't go about calling me one-dimensional.. you're the idiot the
>recreates the same report by hand every week LOL

I don't generate any reports. You seem to need constant reminding about
this. Memory as weak as your intellect?

Or are you so incapable of providing serious and detailed examples of
OLAP/MDX or even SQL that all you can manage is attributing to me the
limited set of Excel skills you're barely capable of?

Quantcast