Simulation Software for Multiple Variable Inputs

Take a look at Monte Carlo simulation. There's tons of software available, some of it is free.

But you are GURU so you already have the answer, No?

Seriously, I can only think of Neural Network programming software that may be able to do the job, but i can not think of how the training has to be done - such software have to be trained to learn the decision making process.

You may have to have someone develop one for you.

Thinking Softwares are very tedious to develop, and the Neural Networking Programming software requires the developer to be very mathematical because the surfaces are often hilly and must be extensively tested. Besides, you have to define the type of probabilistic situation you are thinking of operating in so that the developer can tailor the program to think in that realm.

I hope that this helps.

What's wrong with excel?

I agree with editorgbanalysts. A Neural net will do what you describe. Aspen has (had) a group that were experts in this. They have set up several chemical plant startup sequences based on a historical startups performed different ways. They also can setup operating parameters to optimize around one component performance. I have never seen an Applicaton that had hundreds of inputs.

In a pinch one, thing to try is Northwest Analytical Quality analyst. I have used a free trial and was very satisfied. It includes a very simple and easy to use regression analysis that can take up to 7 independent variables and predict the outcome of the dependent variable. In my application we had about 50 independent variables and did them seven at a time. Picked the most significant of each and reran using the top of each. the results were both surprising and very effective predicting the outcome based on the few significant variables. (no I don't work for NWA)

If its only based on statistic, Excel will do.

It depends on what kind of "probability" you're looking for. You can test fit the data to see witch "formula" or "method" give best result then use that to predict future result.

I would be very interested in using excel since most of the data I use is in an excel file and cut and pasted to a statistical package.

I have never been able to use the regression tool in excel for any more than one variable at a time. It would be very tedious to do 50 without a multiple regression routine to at least determine the significant actors. I have also not been able to plot a 95% (or any) confidence curves on a predictive plot from excel.

Is there a way to do a 7 or 10 variable regression?

Have you written a routine that you would share with us?

Thanks,

Paddler

Warning: I know nothing about statistics, so, I'm not sure if this helps.

But, I manages to extract m1, m2 and c from data created using [y1..y100] = m1[x1..x100] + m2[z1..z100] +c with a bit of randomness thrown in.

by following this article:-

http://office.microsoft.com/en-us/excel/HA011119631033.aspx?pid=CH062528011033

The only difficult bit was entering the LINEST function: you have to select three cells then type the function =LINEST(yarray,xarray,TRUE,FALSE) in the formula bar, then press Ctrl/Shift/Enter.

(I will almost certainly use this to smooth out measured results from electronic circuits.)

hai

We are ERP software developers. We have the solution for your requirement. For further details please contact Sasikala 050 4221864