| Re: Fitting curve to data [message #93632 is a reply to message #93631] |
Wed, 14 September 2016 08:06   |
wlandsman
Messages: 743
Registered: June 2000
|
Senior Member |
|
|
On Wednesday, September 14, 2016 at 10:23:09 AM UTC-4, liam....@gmx.co.uk wrote:
> Hi all,
>
> I am trying to it a curve to some data points, and have managed to get confused with all the interpolate and spline IDL routines (and more importantly, can't seem to get any of them to work).
>
> As an example, I have a range of depths (in cm) and a value of something at each depth, e.g.
>
> depth = [0.01, 0.1, 0.4, 0.9, 3.5, 14.5, 57.9, 231.7, 926.8]
> value = [2.1, 2.1, 1.9, 1.8, 1.3, 1.0, 0.8, 0.5, 0.4]
>
> What I want to do is fit a curve to the data as best I can (the curve doesn't have to go through each point exactly), on a finer set of depths, e.g.
>
> interp_depth = findgen(1001)
>
> I've tried things like:
>
> interp_val = spline(value,depth,interp_depth)
> interp_val = interpol(value,depth,interp_depth,/spline)
>
> Can anyone offer any help? Is there some sort of cubic least-squares routine or something I should use? I can get linear interpolation to work, but was wanting something more 'realistic'.
>
I think you will get much better results by recognizing that your X axis is logarithmic and doing something like
interp_depth = findgen(1001)+0.1
y = interpol( value,alog10(depth), alog10(interp_depth),/spline)
Of course you need to be very careful when extrapolating outside of the supplied X range. --Wayne
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
