File APIs for Java Developers
Manipulate DOC, XLS, PPT, PDF and many others from your application.
The moose likes Meaningless Drivel and the fly likes Question on statistical analysis Big Moose Saloon
  Search | Java FAQ | Recent Topics | Flagged Topics | Hot Topics | Zero Replies
Register / Login
JavaRanch » Java Forums » Other » Meaningless Drivel
Bookmark "Question on statistical analysis" Watch "Question on statistical analysis" New topic

Question on statistical analysis

Ann Basso
Ranch Hand

Joined: Jul 26, 2007
Posts: 48
Hi all! I have a question regarding statistical analysis and was wondering if anybody can provide any pointers.

I want to compare two methods/approaches of building a structure and find out which approach is better.

There are 10 parameters, for which I have collected data. For example, lets say one parameter is Time To Cure (TTC). I have 100 values for this "Time To Cure" parameter for each of the methods. Another parameter is Strength and so on.

The methods have to be compared based on these parameters. Every parameter is different. For example, if TTC is less for Method 1, then the method 1 is better. But if Strength is less in Method 1, then Method 2 is better and so on.

My approach is to assign weightage to each parameter and basically compute some score. But I am not sure how and what statistical functions to apply that can provide some scientific basis to the conclusion. Also, I want to know if there are enough values to make that conclusion (something to do with Significance). Or which parameter values are useless to arrive at the conclusion (if Std Dev. of some parameter is too much then may be it is statistically no good?).

Is there a standard way to do this kind of thing or are there some standard numbers that should be produced so that other can understand and appreciate the study?

thank you!

marc weber

Joined: Aug 31, 2004
Posts: 11343

In comparing each of the like parameters (for example, Method 1's TTC vs. Method 2's TCC), you want a way to decide whether your sample really shows one method is better, or whether the difference should be attributed to normal variations. I think you're looking for a t-test.

However, when it comes to weighting different parameters (for example, how much you might value TTC vs. Strength)... That seems subjective.

"We're kind of on the level of crossword puzzle writers... And no one ever goes to them and gives them an award." ~Joe Strummer
I agree. Here's the link:
subject: Question on statistical analysis
It's not a secret anymore!