Split vector in groups of data and plot best fit

Good evening,

I have a group of data in the following format:

    Dat = [ NaN ; NaN ; 23 ; 34 ; 5 ; 12 ; 4 ; 123 ; NaN ; NaN ; Nan ; 23 ; 4 ; 1 ; 4 ; 21 ; Nan ; 234 ; 12 ; 3 ; 12 ; NaN ; 342 ; 32 ; NaN ; NaN ; NaN ; NaN ; 34 ] ;

I would like to take the groups of values between the NaN sets and plot best fit equation. So I would take:

a = [23 34 5 12 4 123 ] b= [23 4 1 4 21] c= etc....

and plot a best fit curve for each of these.

I have managed to come up with a way using indices and a for loop, but it is way to complicated. Also this results to some bugs coming up as we saw with my teacher. So is there any way you would suggest?

KR,

Kosta

[Konstantinos Tsitsilonis]

To Clarify, below is the code:

      %Condition Matrix = [n x 2] matrix of the format as described above, look 'Dat'. 

       %Find the number of voyages

       for i = 1:2 %number of loading conditions i.e. ballasted and laden

            finite_indx = double(~isnan(Condition_Matrix(:,i))) ; 
            shifted_indx = [0 ; finite_indx(1:end-1) ] ;
            multiplied_indx = finite_indx .* shifted_indx ;

            Logical_Matrix = [ Condition_Matrix(:,i) , finite_indx , shifted_indx , multiplied_indx ] ;

         Voyages_indx = sum( double(Logical_Matrix(:,2:end) == [1 0 0]),2) == 3  ; 
         Voyages = sum(Voyages_indx) ;         

         %Indexes
            %Index of first data recording value of voyage 
            idx_low = find(sum( double(Logical_Matrix(:,2:end) == [1 0 0]),2) == 3) ;
            %Index of last data recording value of voyage
               idx_int = sum( double(Logical_Matrix(:,2:end) == [0 1 0]),2) == 3 ;
            idx_high = find([idx_int(2:end) ; true]);

             for j=1:Voyages

                 %Power vs. RPM Matrix of each voyage 
                 P_mat = [P_b(idx_low(j):idx_high(j)) , RPM(idx_low(j):idx_high(j)) , Speed(idx_low(j):idx_high(j))]  ;

                 p(j,:)={P_mat} ; 
             end   
       end

Answer 1

[Mudambi Srivatsa]

You can use the following code to split the vector in groups of data:

Dat = [ NaN ; NaN ; 23 ; 34 ; 5 ; 12 ; 4 ; 123 ; NaN ; NaN ; NaN ; 23 ; 4 ; 1 ; 4 ; 21 ; NaN ; 234 ; 12 ; 3 ; 12 ; NaN ; 342 ; 32 ; NaN ; NaN ; NaN ; NaN ; 34 ] ;

count = length(Dat);

% get the indices of NaN elements
nan_indices = find(isnan(Dat));

groups = [];

% pick the elements between NaN elements
for i = 1:length(nan_indices) - 1

      % range to pick the elements 
      first =  nan_indices(i) + 1;
      last = nan_indices(i+1) - 1;    

      % pick only if there is atleast one element between NaN values
      if( last > first)        
          groups = [groups {Dat(first:last)}];
      end
  end

last_nan = nan_indices(end);

% add elements that are after the last NaN
if(count > last_nan)
    groups = [groups {Dat(last_nan+1:end)}]
end

The "groups" will have chunks of data between NaNs. In this case, it will have 5 cell arrays. You can access them and use as follows:

a = groups{1}
b = groups{2} and so on..

[Konstantinos Tsitsilonis]

Thanks a lot!