# Thread: What statistics formulas can be used?

1. ## What statistics formulas can be used?

I want to know what formula can I use to get an average (total?) error. What I did was try a method on 10 different samples a single time (obtaining 10 values), and then I tried an alternative method on those 10 samples (obtaining 10 new values). I then compared each of the values I obtained between corresponding samples (with the percentage error formula) and obtained 10 values which I considered the errors (I want to highlight that I did not perform repeated tests to obtain each of the errors of the method, it was single test for each method on each sample). So if I have ten error values which range from say -5% to 6% and I want to know what the average range of error is obtained form those values what formula do I use? I am assuming that saying my error is from -5% to 6% is wrong so I want to know if there is a way to convert those values into something like +/-4%, if it is even possible because maybe there is no formula for that? I was thinking maybe standard deviation or standard error, but I think those are only used when comparing means, which is not my case.

2. Sorry if I am not being clear, I don't have a very good knowledge of this sort of thing. Here is what I hope a more detailed explanation of what I did and what I want to know:

I used two programs to count certain objects present in digital photographs. The objects had specific characteristics which the programs had to identify to count the objects. For this reason when I ran each of the 10 photographs in the program each program gave me different values. I know that one program is more accurate than the other. I then calculated the percentage error of the values I got. I did this by substracting the value from the "accurate" program from the value of the "innacurate" program. I did this for each of the photos. Then I divided these values by the values of the "accurate" program and multiplied by 100. So this is I believe the percentage error (if I am correct that is the name right?). So now I have a list of 10 "percentage errors" which range from -5% to 6%. So if I have to describe my results, do I say the error obtained from the "innacurate" program was -5% to 6%? Or is there a way to convert that to say +/-4%?