How to run a function

Hello guys, I'm having some problems to how use this function. This function make a data fit to a binomial distribution.

function [sse, curve_x, curve_y] = binomfun(params, data_x, data_y)

    N = params(1);
    p = params(2);
    n = params(3)

    f0 = binopdf(0:N-1, N, p);
    F0 = binocdf(0:N-1, N, p);
    curve_y = n.*f0.*(1-F0).^(n-1);

    data_points = length(data_x)*N;
    new_x_range = 0:1/(data_points-1):1;
    curve_x = 0:1/(N-1):1;
    curve_y_interp = interp1(curve_x, curve_y, new_x_range);
    data_y_interp = interp1(data_x, data_y, new_x_range);

    if max(curve_y_interp) == 0
        scale = 1;
    else 
        scale = max(data_y_interp)/max(curve_y_interp);
    end
    curve_y_interp = scale*curve_y_interp;
    curve_y = scale*curve_y;

    error_vector = curve_y_interp - data_y_interp;
    sse = sum(error_vector .^2);

 function params_min = find_min(data_x, data_y, N_range, p_range, n_range)

    sse_min = realmax;
    params_min = [0,0,0];
    for N = N_range
        for p = p_range
            for n = n_range
                params = [N, p, n];
                [sse, curve] = binomfun(params, data_x, data_y);
                if sse < sse_min
                    sse_min = sse;
                    params_min = params;
                end
            end
        end
    end

I don't know who is data_x. I know that data_y is my data, my data have 668700 points between 0 and 1. The problem always is at interpolation part. But, the question is: Who is my input data scale (data_x)?

Answer 1

[Walter Roberson]

What does your y(1) the value of ? Is it associated with, for example, a particular time or temperature? Is your y just a sequence of data, first point, second point, and so on? Are they equal intervals apart?

If you need to, you can use x = 1:length(y)

[ViniAlmeida]

it's the first point, second, but for example, in my data I have a lot points tha is the same. The point 45 is the same of 300 that is equal to 0.5. When i tried x = 1:length(y) and run, the error is

Error using griddedInterpolant
Interpolation requires at least two sample points in each dimension.

Error in interp1>Interp1D (line 346)
   F = griddedInterpolant(Xext,V,method);

Error in interp1 (line 241)
   Vq = Interp1D(X,V,Xq,method);

Error in binomfun (line 15)
    data_y_interp = interp1(data_x, data_y, new_x_range);

[Walter Roberson]

Your data_x and data_y are scalars it appears.

[Walter Roberson]

It is not a problem for two points to have the same y value provided they have different x values. It is a problem for two points to have the same x values. If your point 45 refers to the same location as your point 300 then you cannot use interp1 (but you might be able to use interp2 in some cases.)

[ViniAlmeida]

 function params_min = find_min(data_x, data_y, N_range, p_range, n_range)

I see now that I use this fuction and this function call the other binonfum, now, the problem is the interpolation. Because I use:

N_range = 50:250
p_range = 0.4:0.001:0.6
n_range = 1:4
data_y = vector 668700x1
data_x = data scale (I don't know)

how is data_X? pleease, This don't woork :((

[Walter Roberson]

data_x = (1:size(data_y,1)).';