How to reshape 1-D to 2-D matrix with missing data?

I need to reshape a 1-D array into a 2-D matrix. The difficulty is that there is random missing data so the RESHAPE function cannot be used in the typical fashion.

I have a set of 3 arrays: x-coordinates, y-coordinates, and a scalar.

For example, this is my input:

x = [ 1 1 1 1 1 2 2 3 3 3 ]

y = [ 1 2 3 4 5 1 2 1 2 5 ]

c = [ 2 1 2 1 2 1 2 1 2 1 ]

and this is my desired output:

x = [ 1 1 1 1 1 ; 2 2 2 2 2 ; 3 3 3 3 3 ]

y = [ 1 2 3 4 5 ; 1 2 3 4 5 ; 1 2 3 4 5 ]

c = [ 2 1 2 1 2 ; 1 2 NaN NaN NaN ; 1 2 NaN NaN 1 ]

Help is much appreciated.

[Azzi Abdelmalek]

This is not clear

[dpb]

In general something like this is tough as one presumes the location of missing values is also more or less random. For any give in case it's pretty much trivial; generically to discover the pattern of the missing values is likely more difficult.

In your case above, one can presume from the output that the first row is complete -- is that a fair assumption in general, however?

Answer 1

[Simon]

I figured out the solution.

First I find the grid vectors:

xgv = sort(unique(x));

ygv = sort(unique(y));

Then I create a mesh grid:

[X,Y] = meshgrid(xgv,ygv);

Finally, I use an interpolation function, but no interpolation occurs because the points are given:

C = griddata(x,y,C,X,Y);

Thanks for your input ya'll

[Lev Dun]

Oh man great solution! I thought I'd have to loop it :P

Answer 2

[Joseph Cheng]

Well you could do something like this.

x = [ 1 1 1 1 1 2 2 3 3 3 ]
y = [ 1 2 3 4 5 1 2 1 2 5 ]
c = [ 2 1 2 1 2 1 2 1 2 1 ]

X = [min(x):max(x)]'; %since x is rows you know you have a maximum of X rows.
X = repmat(X,1,max(y)); %then make it a block based on maximum y value.  
Y = [min(y):max(y)];
Y = repmat(Y,max(x),1);
C = nan*zeros(max(x),max(y));

for ind = 1:length(c)
    C(x(ind),y(ind))=c(ind);    
end

where by building the X and Y based on the knowledge of the structure and a C full of NaNs. Then use the x and y positioning to know which row and column to put the corresponding c value. You may be able to get away from the for loop but forgot how to get C(x,y)=c type of syntax to work out. Something isn't clicking on why that wouldn't work out. (I know matlab is doing the permutations between x and y not x(1) and y(1) then x(2) and y(2) as i'm doing in the for loop.) forgot how to accomplish that without doing a for loop.

[Simon]

Thanks Joseph, I think your method works. I figured out a different system which works non-integer coordinates. See my accepted answer.