[R] Histogram

Sarah Goslee sarah.goslee at gmail.com
Thu Jul 5 22:08:48 CEST 2012


Which gives Jim two ways to arrive at exactly the same result, just
different means of specifying the probs for quantile().

Sarah

On Thu, Jul 5, 2012 at 4:01 PM, Rui Barradas <ruipbarradas at sapo.pt> wrote:
> Hello,
>
> With the confusion between bin size and width the OP started, I'll repost my
> answer with a final line. Sorry for the repetition.
>
>
> h <- hist(x, breaks=quantile(x, probs=seq(0, 1, by=1/20)))
> h$counts
> [1] 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 50
>
>
> Hope this helps,
>
> Rui Barradas
>
> Em 05-07-2012 20:47, Sarah Goslee escreveu:
>>
>> There's no reason you can't do that with normally-distributed data,
>> though I'm not sure why you'd want to. My point was rather that you
>> can't specify the bin width and size both. If you let the bin size
>> vary, this will work:
>>
>> set.seed(1234)
>> mydata <- rnorm(1000, mean = 2, sd = 4)
>> mydata.hist <- hist(mydata, breaks=quantile(mydata, probs=seq(0, 1,
>> length.out = length(mydata)/50 + 1)))
>> mydata.hist$counts
>>
>>
>> Sarah
>>
>> On Thu, Jul 5, 2012 at 3:37 PM, Jim Silverton <jim.silverton at gmail.com>
>> wrote:
>>>
>>> Thanks Sarah!!
>>> Ok so if I have say x = runif(1000,0,1) say instead if the normal and I
>>> want
>>> a histogram with bins that have an equal number of observations. For
>>> example
>>> if I want each bin to have 50 observations, how do I do this?
>>>
>>>
>>>
>>> On Thu, Jul 5, 2012 at 3:34 PM, Sarah Goslee <sarah.goslee at gmail.com>
>>> wrote:
>>>>
>>>>
>>>> Hi Jim,
>>>>
>>>> You can't specify both number of bins and bin size. You can specify
>>>> breaks: either the number of bins or the location of breakpoints. A
>>>> histogram with 20 bins of 50 observations each must by definition come
>>>> from a uniform distribution.
>>>>
>>>> What are you trying to accomplish?
>>>>
>>>> Sarah
>>>>
>>>> On Thu, Jul 5, 2012 at 3:29 PM, Jim Silverton <jim.silverton at gmail.com>
>>>> wrote:
>>>>>
>>>>> I have a column of 1000 datapoints from the normal distribution with
>>>>> mean 2
>>>>> and variance 4. How can I get a histogram of these observations with 20
>>>>> bins with each bin having 50 observations?
>>>>>
>>>>> --
>>>>> Thanks,
>>>>> Jim.
>>>>
>>>>
>>
>



More information about the R-help mailing list