CODEWITHSUNDEEP
javascriptmathcurve-fitting
zdebruine
zdebruine

Reputation: 3807

JS exponential curve fit

I'm trying to make a simple function to do an exponential curve fit given a dataset. I've written the following:

  function expCurveFit(d){
    var sum_x2 = 0, sum_lny = 0, sum_x = 0, sum_xlny = 0, n = d.length;
    for(var i=0;i<d.length;i++){
      var x = d[i][0];
      var y = d[i][1];
      sum_x2 += x^2, sum_lny += Math.log(y), sum_x += x, sum_xlny += x*Math.log(y);
    }
    var a = ((sum_lny*sum_x2)-(sum_x*sum_xlny))/((n*sum_x2)-sum_x^2);
    var b = ((n*sum_xlny)-(sum_x*sum_lny))/((n*sum_x2)-sum_x^2);
    return [a,b];
  }

but a and b don't seem to be correct. I'm working from the basic formula at Wolfram. The inputted array is in the form [[x1,y1],[x2,y2]...[xn,yn]]. Anyone have working suggestions, plug-ins, libraries, or revised functions?

Upvotes: 2

Views: 2208

Answers (2)

Rafael Miller
Rafael Miller

Reputation: 335

Besides the XOR operator as marked as the answer for this question, there's a mistake on the a and b coefficients.

The curve that you will fit using this method is Y = B * exp(A * x), as is written in Wolfram, but B = b and A = exp(a)!

So you have to change the return [a,b] to [ Math.exp(a), b] (if you want to use this array as a "equation")

Upvotes: 1

Paul
Paul

Reputation: 141827

^ is the XOR operator in JavaScript, not exponentiation.

Replace x^2 with Math.pow(x,2) or x*x and do the same for both places where you have sum_x^2.

Upvotes: 5

Related Questions