It seems to me you don't really understand what it means to define a parameter. If you don't define something as one of the analysis parameters it won't be analysed; it's as simple as that.

__________________
Cheers,
Paul Edstein
[Fmr MS MVP - Word]

I can already define parameter inside BW. If I would be defining parameters in tool I am asking for and still waiting for accurate reply/answer, no time would be saved.

NewPuppy, I apologize for the delay; I was away for a while and didn't realize this conversation was still going on. If you like, we can try to discuss this a bit more. Paul may be right, but while I read your 12-19 04:00 post (and reread, and rereread it) I come to suspect that maybe the problem we're having is in a difference of understanding about the word "compare".

At one point you said you might (for example) need to "compare country vs. age", and went on to name a bunch of examples: "europe vs between 20 and 30...asia vs between 20 and 30...germany vs 20" and so on. Now, in statistical analysis there's a thing called "cross-tabulation", and it sounds to me as though this is what you're talking about. For example, if I have 8000 responses to a survey, I can ask my program to "cross-tabulate" the positive responses to question #21 by age and race, and it'll create a table something like this:Is this the kind of "comparison" you had in mind? If so, then we probably misunderstood you at first—I did, anyway—and Paul may be able to suggest something. Me, for this kind of analysis I'd look at SPSS or SAS or one of the other packages that are designed for this kind of analysis.

Bob Thank you for reply. You asked me if your example if what I meant. No, not at all. The reason for this is because all such parameters can already be compared inside BW (excel's addon). As soon as I am required to choose WHICH parameters must be compared (parameter1 vs parameter2 OR filter1 vs. filter2 etc) then this is pointless because NO time is saved. I can already do everything like this inside BW and much more.

The main feature of the tool I am asking for is possibility that I don't need to define WHAT to compare because everything must be compared:

- every filter with every other filter (level one of comparison)
- every filter's parameter with every other filter's parameter (level two)
- every filter's parameter's value with every other filter's parameter's value (level three)

even the filters/parameters that are NOT enabled for view of table must also be compared. This BW addon, not just it is isolated in excel but also very customized.

The goal of having such tool I am asking for is simply defining only WHAT counts as being out of order. Thats all. However I would expect from tool some reports where I wouldn't need to define numerical value: ''what counts as being out of order'' for example (only random chosen example) in report of the tool I would like to see answer on:

''Why are being too many newcommers refused in particular company with job applications?''

I wanted to show in YOUR example (table) what would I define in the tool I am looking for but you are showing only 2 filters and 8 parameters. Impossible to provide case on your example - im having few thousands filters and few tens of thousands of parameters. If I try to anyway, just to be as much as possible understandable, I would define in the tool (in your example) something like (note: with your comment ''positive responses'' I am assuming something like well learned lecture related to the lessons topic number 21 as you said):

''What is causing asian people from 36 to 50 years old to have so many troubles in understanding the lectures on topic 21 comparing to asian people from 18 to 21 years old - could be wrong books used? could be they missing during the class? why would they ignore the class? why 18 to 21 years old attended the class then? etc''

But all those answers can be gathered obviously only from filters (enabled or not, doesn't matter). So for the latest question in quoted text there should be something like:

filter: class attending (further directly related to your two filters)
parameter1: reason/goal of attending the lecture
parameter2: interest for attending the lecture

Hopefully now im little more understandable. Once again: In the tool I am looking for I should only need to define:

If numerical definition needed: Define the boundaries
If numerical definition NOT needed: Defining what kind of report I want

BUT I definitely don't want to define what filters/parameters to compare. Otherwise for sure no time is saved.

Ok, so that idea wasn't right. But I still want to understand this, so maybe I should look more closely at your example and ask for details about it. In your example, the question you want answered is "'With regard to company X, why are too many applicants being refused employment?". I infer that you have a database of job applications to many companies, and you see that for many companies there's an average rate of refusal vs acceptance; but in one particular company you see that the percentages are far from the average, and you want your analysis to discover why. You hope that by examining all the data about these companies and these job applications, you might be able to spot a relationship between vectors, or a combination of relationships, that will make the matter clear. Am I right about that? So you're hoping for a product that will search through the many, many thousands of data looking for a pattern, or rather a departure from the pattern, that may explain it?

It was just an example. I could and should get thousands of different reports from automated tools I am asking for but obviously every piece or report should contain (and it does) elements/parameters/filters/values in the BW (Business Warehouse) - very customized Excel's addon.

Yes you are right - I would like to get reports with method ''comparing everything to everything else'' without needing to define what should be compared. As soon as I define, time is completely wasted because I could define in BW addon already. The point is to NOT need to waste time and define parameters AND (!!!) draw charts - everything needs to be automated. This is the tool I am searching for. I should only be required to define what I am interested for and NOT what to compare.

Answer to your last sentence in your latest message: yes

Well, I may be wrong, but I think you're hoping to find a program that will think for you. Programs, unfortunately, don't think; they just calculate, once someone explains to them how to do it. You would have to define at least part of the problem for the program.

Put it this way: If you were to write such an add-on yourself, what (exactly) would you tell it to do? You can define to your program what constitutes a data vector (age, nationality, profession, whatever), and tell the program to take the vectors in every possible combination of pairs. But here's the question: Once your program is looking at two vectors—educational level and religion, perhaps—what should your program do with them?

So far we've just said "compare", but for a program to be able to follow instructions we have to be more specific. What about the data are you looking at? What would you do to advance your analysis?

I've got a notion that despite your insistence that you want to compare "everything to everything", what you have is a particular datum—a bunch of people who answered 'yes' to question 13, for example—and you want to find a correlation between that answer and one or a combination of demographic vectors. A really exhaustive program, therefore, would calculate the correlation between question 13 and each demographic vector; then between question 13 and each combination of two demographic vectors; then between question 13 and each combination of three demographic vectors; and so on. Then it would list the correlations in descending order of absolute value.

If so, you would have to define to your program at a minimum a) which are the demographic vectors and b) which question(s) on the survey you want analyzed in this way. But I'm not at all convinced that this is what you want, still.