Least_mean Squares

I am trying to write the below function.

for(ct = 1:N )

sum_xy = sum_xy + x(ct)*y(ct) ;

sum_xx = sum_xx + x(ct)*x(ct) ;

sum_x= sum_x+x(ct)

sum_y = sum_y+y(ct)

end

% Compute best value of b

%b_tilde = sum_xy / sum_xx ;

b_tilde = (sum_xy - ((1/N)*sum_x*sum_y))/(sum_xx-((1/N)*sum_xx));

fprintf(1,'b_tilde is %1.10f \n',b_tilde) ;

I havent done a_tilde yet but i presume its (sum_y/N) - b_tilde*(sum_x/N)

any advice appreciated.

Answer 1

[Star Strider]

To make your code simpler and more efficient use the sum and mean functions. You can likely avoid the (explicit) loops.

Since you will likely need to do element-wise multiplication, also see Array vs.Matrix Operations.

[Jason Earls]

yes because I dont know if i am writing out the above equation correctly

[Star Strider]

The way to test that is to produce some data with a known slope and intercept, and see if your code reproduces those parameters.

Meanwhile use the sum and mean functions to make your ocde easier to write and execute.