How to interpolate arrays for smoothening?

Hi all,

I'd like to interpolate an array of x-values to apply a spine function to it later (I'd like to smoothen my dataset). However, the x-array is a discontinuous vector, so the general explanation how to interpolate with interp1 does not work for my data. This is the array:

14.1 13.7 12.8 11 9.22 6.18 5.57 5 4.27 3.42 2.57

Does any of you have an idea how I could interpolate this array? Or maybe other ideas how to smoothen it, if not?

Thanks!

[Adam]

Do you not have a sampling vector to go with that? i.e. the values at which those results were calculated. If you know where your discontinuities are can you not just use interp1 piece-wise between discontinuities?

Answer 1

[Andrew Reibold]

To interpolate an array, you need the following

X: An array of data, lets call it your independant data.

Y: An array of data corresponding to X. For each value of X there is a value for Y.

X_interpolated: An array of data of NEW X points at positions you want Y interpolated at.

Example:

X = [1 2 3 4 5 6 7]
Y = [10 11.5 14 15 16.5 18 20]
X_interpolated = [1:.5:7]

Y_interpolated = interp1(X,Y,X_interpolated)

Now if you plot

plot(X_interpolated,Y_interpolated)

you will find that points have been interpolated to fill in the gaps of the initial data. Specifically, these points are located at whatever you chose for X_interpolated

If you want to make it curvy or smooth after that, if appropriate, you could use 'smooth' to take a moving line average of the data like this

Y_smooth = smooth(Y_interpolated)
plot(X_interpolated, Y_smooth)

Otherwise I would suggest trying to use something like polyfit for your function... if its a function.

[Ellen]

Thanks Andrew for your response. The problem with my data is that it is not really a function, but a set of scatterpoints which do not relate. I've checked a lot, including your suggestion for polyfit, but all of this does not work for my set of data. Thanks anyway though!