Need help for a statistical tool for my test

[mark.ramos]

Hi!

I am a software programmer and I would like to make a performance analysis of my software. One of the objectives that we need to achieve is to create a module for accurately computing the payment. Now, my expected result is that all computations are accurate. What statistical tool should I use to support my observations. I am confused in using a chi-square test for this. The expected outcomes are either Accurate or Not Accurate. Since I assume that my computations are accurate, the frequency that I set for accurate is 10 and 0 for not accurate in the case of 10 trials. I hope someone can help me on this one.

Thank you!

 

[NoMoreExams]

So you have a binomial distribution?

 

[CRGreathouse]

What information do you have? How do you tell if a calculation was accurate or not?

 

[mark.ramos]

What information do you have? How do you tell if a calculation was accurate or not?

Well, to test if it is accurate, I first compute it manually (without the system) then I compare the computed value of the system and if it matches my manual computation I consider it accurate. By the way I would like to know what statistical test should I use if I want to determine if a module is working or not. Like I have an objective that I should create a module for generating reports, and I have to make a test that will support that I have achieved my objectives. What test should I use?

Thank you!

 

[gumi_kr]

So you have 2 sets of data:

1) calculations from your program
2) handmade calculations

and you want to check it's accuracy, right?

My thoughts: (i'm not advanced in statistics though):
1) define what's 'accurate' (5% error?)
2) If you want to then define 'Confidence interval' and do some basic test or as you have written do some 'Goodness of fit' test (you can support it with p-value test, but it will probably give you the information: 'your hipothesis cannot be rejected' ...)

but let's wait till somebody wiser give you better answer;)