Re: PDF of Time Series Data
- From: roberson@xxxxxxxxxxxxxxxxxx (Walter Roberson)
- Date: Thu, 24 Apr 2008 21:01:24 +0000 (UTC)
In article <fuqr1d$b83$1@xxxxxxxxxxxxxxxxxx>,
Prantik Mazumder <mazumderp@xxxxxxxxxxx> wrote:
I need to calculate the probability density distribution
of a measured time-series data. While matlab offers
probability distribution functions (pdf, cdf etc.) to
generate calculated time-series data, I could not find
anything that will do the reverse, that is read the
measured time-series data and then generate the PDF
distribution. Any suggestion would be appreciated.
Perhaps I am misunderstanding the problem, but isn't that just
hist() or histc() with sufficiently many bins, with the count
for the bin being the probability weight for the discrete interval
represented by that bin?
You cannot calculate an exact -continuous- pdf from a finite number of
samples. You -can- (ks-test) make calculations to determine the
probability that a set of samples are drawn from a particular
known distribution.
--
"All human knowledge takes the form of interpretation."
-- Walter Benjamin
.
- References:
- PDF of Time Series Data
- From: Prantik Mazumder
- PDF of Time Series Data
- Prev by Date: Re: coeffiecients of the begining and of the end of graph
- Next by Date: Re: reshaping cost
- Previous by thread: PDF of Time Series Data
- Next by thread: Re: PDF of Time Series Data
- Index(es):
Relevant Pages
|