Issue with rounding of smoothing parameter in fit

I am trying to use the fit() function to produce smoothed spline curves, and display multiple curves on one plot for comparison. I used the Curve Fitting app to determine the best smoothing param based on relative smoothness and fit stats

s_param = [0.99985344]
spartcurve = fit(x_elev,y_spart,'smoothingspline')
    spartop = fitoptions(spartcurve)
    spartop.SmoothingParam = s_param

Everything works smoothly, except Matlab seems to be rounding my smoothing parameter to 1, which is a problem because small numerical changes produce large changes in the roughness/smoothness of the curve. The console shows the following after running the above:

    spartop =

         Normalize: 'off'
           Exclude: []
           Weights: []
            Method: 'SmoothingSpline'
    SmoothingParam: 1.0000

This seems like something that might be related to how Matlab handles floating point numbers but I am not well versed enough to figure it out.

Thank you for your help!

[John D'Errico]

So, what does spartop.SmoothingParam return? Don't just copy the 5 digits that you see.

MATLAB does NOT round anything to 1, except when it displays the number. The number is still stored in double precision. Hint:

format long g

Anyway, if your problem is that sensitive to tiny variations on the order of 1 part in 1e4, then you have a problem with what you are trying to do.

[Jacob Simon]

To my knowledge, what I included originally is what spartop.SmoothingParam returns; viewing it in the object window shows the same values.

Thank you for your help, I wasn't aware of the format options! That kept the parameter value precise, although the plotted curve is not changing to reflect that... I believe you maybe right in saying that the process is flawed so i might just be at an impasse here

[Mathieu NOE]

well

in case that could be of any help , this is a simple code to do some data (experimental or model) smoothing

might be enough for simple tasks and you have only one parameter to dela with

regards,