| Calculating cumulative probability using cgHistoPlot [message #78261] |
Tue, 08 November 2011 10:10  |
Xin Tao
Messages: 40
Registered: April 2011
|
Member |
|
|
Hi all,
I'm using cghistoplot, and I'm confused by the cumulative probability
calculated by cghistoplot. I'm wondering whether any one can give me
some help here.
Suppose we have histdata like this: [2, 0 , 0, 3, 5, 4], then if
we use the way of cghistoplot to calculate the cumulative probability
like this:
cumTotal = Total(histData, /CUMULATIVE) ;; gives us
[2.00000, 2.00000, 2.00000, 5.00000, 10.0000,
14.0000]
probability = Scale_Vector(cumTotal, 0, 1) ;; gives us
probability = [0.00000 0.00000 0.00000 0.250000
0.666667 1.00000]
This is kind of counter-intuitive to me, because the first value of
histdata is clearly 2, but the probability is 0 until the 4th value.
However, I'm not experienced in data anlaysis, and I might have
misunderstood something about "cumulative probability" here.
It seems to be more natural to me to define the cumulative probability
in the following way:
probability = total(double(hist)/total(double(hist)), /cumula)
Am I right?
Xin Tao
|
|
|
|
| Re: Calculating cumulative probability using cgHistoPlot [message #78350 is a reply to message #78261] |
Tue, 08 November 2011 14:43  |
Xin Tao
Messages: 40
Registered: April 2011
|
Member |
|
|
Yes, that's updated. My web browser probably remembered the page and
displayed the old version. I had to manually refresh it to get the
newer version. :)
thanks again,
Xin Tao
On Nov 8, 2:34 pm, David Fanning <n...@dfanning.com> wrote:
> Xin Tao writes:
>> Is the link right? I seem to get the same results.
>
> Yes, the link is right, but I don't get anything like
> the same result. At least for your test data. I would
> hope the result is at least *similar* with other data
> sets.
>
> Are you sure you are running the new version? ;-)
>
> Cheers,
>
> David
>
> --
> David Fanning, Ph.D.
> Fanning Software Consulting, Inc.
> Coyote's Guide to IDL Programming:http://www.idlcoyote.com/
> Sepore ma de ni thui. ("Perhaps thou speakest truth.")
|
|
|
|
| Re: Calculating cumulative probability using cgHistoPlot [message #78351 is a reply to message #78261] |
Tue, 08 November 2011 14:34 |
David Fanning
Messages: 11724
Registered: August 2001
|
Senior Member |
|
|
Xin Tao writes:
> Is the link right? I seem to get the same results.
Yes, the link is right, but I don't get anything like
the same result. At least for your test data. I would
hope the result is at least *similar* with other data
sets.
Are you sure you are running the new version? ;-)
Cheers,
David
--
David Fanning, Ph.D.
Fanning Software Consulting, Inc.
Coyote's Guide to IDL Programming: http://www.idlcoyote.com/
Sepore ma de ni thui. ("Perhaps thou speakest truth.")
|
|
|
|
| Re: Calculating cumulative probability using cgHistoPlot [message #78355 is a reply to message #78261] |
Tue, 08 November 2011 13:09 |
Xin Tao
Messages: 40
Registered: April 2011
|
Member |
|
|
On Nov 8, 11:57 am, David Fanning <n...@dfanning.com> wrote:
> Xin Tao writes:
>
>> Hi all,
>
>> I'm using cghistoplot, and I'm confused by the cumulative probability
>> calculated by cghistoplot. I'm wondering whether any one can give me
>> some help here.
>
>> Suppose we have histdata like this: [2, 0 , 0, 3, 5, 4], then if
>> we use the way of cghistoplot to calculate the cumulative probability
>> like this:
>
>> cumTotal = Total(histData, /CUMULATIVE) ;; gives us
>> [2.00000, 2.00000, 2.00000, 5.00000, 10.0000,
>> 14.0000]
>> probability = Scale_Vector(cumTotal, 0, 1) ;; gives us
>> probability = [0.00000 0.00000 0.00000 0.250000
>> 0.666667 1.00000]
>
>> This is kind of counter-intuitive to me, because the first value of
>> histdata is clearly 2, but the probability is 0 until the 4th value.
>> However, I'm not experienced in data anlaysis, and I might have
>> misunderstood something about "cumulative probability" here.
>
>> It seems to be more natural to me to define the cumulative probability
>> in the following way:
>
>> probability = total(double(hist)/total(double(hist)), /cumula)
>
>> Am I right?
>
> I think you are right. I was both calculating this incorrectly
> and displaying it incorrectly. I think you will be more pleased
> with the updated program. :-)
>
> http://www.idlcoyote.com/programs/cghistoplot.pro
>
> Thanks for pointing this error out.
>
> Cheers,
>
> David
>
> --
> David Fanning, Ph.D.
> Fanning Software Consulting, Inc.
> Coyote's Guide to IDL Programming:http://www.idlcoyote.com/
> Sepore ma de ni thui. ("Perhaps thou speakest truth.")
Thanks David.
Is the link right? I seem to get the same results.
|
|
|
|
| Re: Calculating cumulative probability using cgHistoPlot [message #78358 is a reply to message #78261] |
Tue, 08 November 2011 11:57 |
David Fanning
Messages: 11724
Registered: August 2001
|
Senior Member |
|
|
Xin Tao writes:
>
> Hi all,
>
> I'm using cghistoplot, and I'm confused by the cumulative probability
> calculated by cghistoplot. I'm wondering whether any one can give me
> some help here.
>
> Suppose we have histdata like this: [2, 0 , 0, 3, 5, 4], then if
> we use the way of cghistoplot to calculate the cumulative probability
> like this:
>
> cumTotal = Total(histData, /CUMULATIVE) ;; gives us
> [2.00000, 2.00000, 2.00000, 5.00000, 10.0000,
> 14.0000]
> probability = Scale_Vector(cumTotal, 0, 1) ;; gives us
> probability = [0.00000 0.00000 0.00000 0.250000
> 0.666667 1.00000]
>
> This is kind of counter-intuitive to me, because the first value of
> histdata is clearly 2, but the probability is 0 until the 4th value.
> However, I'm not experienced in data anlaysis, and I might have
> misunderstood something about "cumulative probability" here.
>
> It seems to be more natural to me to define the cumulative probability
> in the following way:
>
> probability = total(double(hist)/total(double(hist)), /cumula)
>
> Am I right?
I think you are right. I was both calculating this incorrectly
and displaying it incorrectly. I think you will be more pleased
with the updated program. :-)
http://www.idlcoyote.com/programs/cghistoplot.pro
Thanks for pointing this error out.
Cheers,
David
--
David Fanning, Ph.D.
Fanning Software Consulting, Inc.
Coyote's Guide to IDL Programming: http://www.idlcoyote.com/
Sepore ma de ni thui. ("Perhaps thou speakest truth.")
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
