Is griddedinterpolant omits NaN?

Hey all, I want to know when we using griddedinterpolant while we have some NaN in our data set, are they NaNs fills by griddedinterpolant? Or they remain as NaN?

Answer 1

[darova]

You can find out by yourself

[X,Y] = meshgrid(0:10);
Z = 0*X;
Z(5:7,5:7) = nan;
surf(X,Y,Z,'edgecolor','none')

[X1,Y1] = meshgrid(0:0.4:10);
Z1 = griddata(X,Y,Z,X1,Y1);
hold on
surf(X1,Y1,Z1,'facecolor','none')
hold off

[Andrey Yatsunenko]

I'm afraid the answer is incorrect.

Griddata is not a close relative to the griddedInterpolant; griddata (despite its name) deals with scattered data, and only up to 3 dimentions.

Answer 2

[Stephen23]

None of the pure interpolation routines will "omit" NaN values.

NaN values will propagate from input to output, just as they should in any mathematical calculation.

Lets try it now by modifying the first example from the GRIDDEDINTERPOLANT documentation:

x = sort(20*rand(100,1));
v = besselj(0,x);
v(23:50) = NaN; % modified
F = griddedInterpolant(x,v)

F = 
  griddedInterpolant with properties:

            GridVectors: {[100x1 double]}
                 Values: [100x1 double]
                 Method: 'linear'
    ExtrapolationMethod: 'linear'

xq = linspace(0,20,500);
vq = F(xq);
plot(x,v,'ro')
hold on
plot(xq,vq,'.')
legend('Sample Points','Interpolated Values')

If you want to remove NaNs from data then you will need to either:

• pre-process the data to remove the NaNs yourself, or
• use high-level convenience functions e.g. FILLMISSING.