# How to get the required data? [closed]

The input is a bunch of numbers. How can I calculate a threshold that there are 10% of the totally amount of numbers are above this threshold. I think this is equally as to calculate the distribution function, but what I need is the threshold number, not the

-
Calculate in what sense? – mbq Nov 26 '10 at 1:02
because I want to sample those numbers based on how big they are, so I have to find a threshold, I want give the top 10% numbers a higher chance to be selected. – baboonWorksFine Nov 26 '10 at 1:28
Please reformulate the header of the question (more specific, less generic) and complete the last sentence to increase the long-term-value of this question for the site/audience. Thank you :) – steffen Nov 26 '10 at 8:19
I've edited your title, to make it fit with other questions. As you may have noticed, questions are usually comprised of things like (a) a brief description of the problem at hand, (b) what are the available data actually, and (c) ... a motivated and understandable question. All three ingredients seem lacking there, which is the reason why you get downvoted. Would you pls make a little effort toward describing what your data set looks like, and take the time to finish your last sentence (nobody will ever complete it for you)? – chl Nov 27 '10 at 11:56

## closed as not a real question by whuber♦, csgillespieNov 30 '10 at 16:38

It's difficult to tell what is being asked here. This question is ambiguous, vague, incomplete, overly broad, or rhetorical and cannot be reasonably answered in its current form. For help clarifying this question so that it can be reopened, see the FAQ.

I'm not sure if this is really what you are asking for, so please clarify or state otherwise if this isn't what you had in mind. In the mean time, here in an approach to calculate a 10% threshold and then also select the elements of that list which are above that threshold. This is using R.

#Generate list of "random numbers" with normal distribution, mean = 0, sd = 3
x <- rnorm(n = 1000, mean = 0, sd = 3)

#Returns the 90% quantile of the bunch of random numbers above
quantile(x, .9)

#Subset x and only return those which are above the 90% threshold.
x[x > quantile(x, .9)]

-

My understanding of the question is that given a series of numbers, which is the number that has 10% of all the given numbers larger than this number and 90% smaller? If this understanding is correct, a software percentile function (see here) or a decile function (see here) might be what is needed.

-
Feel free to provide links to Wikipedia, but at least provide a sentence or two of explanation in your answer. – Rob Hyndman Nov 26 '10 at 5:00
please try to provide context to answers, not just links. – Gavin Simpson Nov 26 '10 at 17:19
I've +1 because you addressed the above remarks from @Rob and @Gavin. The term 'software' seems, however, misleading here because the links you referred to have nothing to do with a possible instantiation in any statistical software (as @Chase did), but this is no more than a remark. – chl Nov 27 '10 at 12:04