| replacing "NAN" values using interpolation in IDL [message #82789] |
Thu, 17 January 2013 02:02  |
abc
Messages: 46
Registered: March 2011
|
Member |
|
|
Dear all, I have an data array. I have many "NaN" values. I want to replace those "NaN" values by the nearby values either using interpolation. Is there any way to do that in IDL.
Thanks in advance
|
|
|
|
|
|
|
|
|
|
| Re: replacing "NAN" values using interpolation in IDL [message #82954 is a reply to message #82789] |
Tue, 29 January 2013 16:47  |
Dick Jackson
Messages: 347
Registered: August 1998
|
Senior Member |
|
|
Dear idlhelp,
That's even easier, assuming that you mean that your data 'x' values are evenly spaced, and can be considered to be [0, 1, 2, ...]
Building on Jeremy's work:
; sample data
data = [0., !values.f_nan, 10., 15., !values.f_nan, !values.f_nan, 5., 4., 5.]
; figure out where there are NaNs and where the useful data are
; Dick prefers "Finite(data)" to "data EQ data" for finding non-NaN :-)
gooddata = where(Finite(data), ngooddata, comp=baddata, ncomp=nbaddata)
; interpolate at the locations of the bad data using the good data
if nbaddata gt 0 then data[baddata] = interpol(data[gooddata], gooddata, baddata)
Hope this helps!
-Dick
Dick Jackson Software Consulting
Victoria, BC, Canada
www.d-jackson.com
On Monday, January 28, 2013 9:27:36 AM UTC-8, idlhelp wrote:
> On Thursday, January 17, 2013 4:05:11 PM UTC+1, Jeremy Bailin wrote:
>
>> On 1/17/13 4:02 AM, idlhelp wrote:
>>> Dear all, I have an data array. I have many "NaN" values. I want to replace those "NaN" values by the nearby values either using interpolation. Is there any way to do that in IDL.
>>>
>>> Thanks in advance
>
>> Yes, there is.
>>
>> Oh, you want to know how? ;-)
>>
>> I would do something like this, if I had an array of locations x and a
>> data array data:
>>
>> ; sample data
>> x = [1., 2., 3., 5., 7., 7.5, 9., 12., 12.1]
>> data = [0., !values.f_nan, 10., 15., !values.f_nan, !values.f_nan, 5.,
>> 4., 5.]
>>
>> ; figure out where there are NaNs and where the useful data are
>> gooddata = where(data eq data, ngooddata, $
>> comp=baddata, ncomp=nbaddata)
>>
>> ; interpolate at the locations of the bad data using the good data
>> if nbaddata gt 0 then begin
>> data[baddata] = interpol(data[gooddata], x[gooddata], x[baddata])
>> endif
>
>> -Jeremy.
>
> Thanks Jeremy, but the problem I am having is that I don't have any sample data 'x'. I have only 'data' and want to use that 'data' file again for interpolating and replacing the NaN value. Any Idea how to do that.
>
> thanks
|
|
|
|
| Re: replacing "NAN" values using interpolation in IDL [message #83869 is a reply to message #82951] |
Thu, 11 April 2013 07:31 |
Chang Liao
Messages: 5
Registered: December 2012
|
Junior Member |
|
|
On Wednesday, January 30, 2013 4:18:53 AM UTC-5, Carsten Lechte wrote:
> On 30/01/13 03:37, Jeremy Bailin wrote:
>
>> I must admit, the silliness factor is the main reason I use the "data
>
>> EQ data" convention. :-)
>
>
>
> Alas, no silliness without punishment: this method fails on windows
>
> IDL <=5.6, see last entry on
>
> <http://www.idlcoyote.com/math_tips/nans.html>. Fortunately, that
>
> version is so old, that even MY place of employment has a newer one.
>
>
>
>
>
> chl
Yh, I have read that article as well. This is a pretty interesting topic of handling NAN in data.
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
