Fill missing NaN values with Interpolation

I have Matlab 2013, and I want to apply spline interpolation in my data matrix with NaN values. Is there any other method which can do this. As fillmissing functon is not available in lower versions.

[Adam]

doc spline
doc interp1

[Tsehaye Gebreteklie]

knnimpute is prefere than spline. you can write the code like this.

clear all;

clc;

Temp = xlsread(A);

Filldata=knnimpute(Temp);

Answer 1

[KSSV]

a = rand(100,1) ;
% put some NaN's
a(randsample(100,20)) = NaN ;
%%interpolate 
x = 1:length(a) ;
a(isnan(a)) = interp1(x(~isnan(a)),a(~isnan(a)),x(isnan(a))) ;
plot(x,a,'.r') 
hold on
plot(x,a,'b')

[EMMA POLLARD]

@KSSV Is there a way to use the above method but omit larger gaps of data? For example if there is a string longer than 5NaN values could it avoid interpolating these sections?

Thanks

Answer 2

[ZaidiN]

Is this spline interpolation? Secondly this doesnt fill the locations of NaN with interpolated points.

[Pawan Sharma]

This is liner interpolation. To make it spline interpolation, add spine as a method of interpolation

a(isnan(a)) = interp1(x(~isnan(a)),a(~isnan(a)),x(isnan(a)), 'spline') ;

It do replace NaNs with interpolated values.