Gaussian psychometric curve fit to the data

Here is the problem I am working on; this should use erfc x = 1:5; y = [0.97 0.89 0.5 0.08 0.03]; we need to get slope and bias (The bias should be close to 3 and slope should be close to 0.78 for this) I need to use cumulative gaussian anonymous function f(x) = 1/2*erf( (

[Star Strider]

Please complete your Question:

f(x) = 1/2*erf( (  ???

What parameters do you want to estimate?

[Janani Thiyagarajan]

Here is the complete question; it got truncated due to symbols usage Here is the problem I am working on; this should use erfc x = 1:5; y = [0.97 0.89 0.5 0.08 0.03]; we need to get slope and bias (The bias should be close to 3 and slope should be close to 0.78 for this) I need to use cumulative gaussian anonymous function f(x) = 1/2 ∗ erf ( (x - bias) / (slope*sqrt(2)) )

Answer 1

[Star Strider]

This works:

x = 1:5; 
y = [0.97 0.89 0.5 0.08 0.03];

f = @(p,x) 1/2 * erfc ( (x - p(1)) ./ (p(2).*sqrt(2)) );
P = fminsearch(@(p) norm(y - f(p,x)), [1; 1]);

Bias = P(1)
Slope = P(2)

Bias =
    2.9815

Slope =
    0.7784

[Janani Thiyagarajan]

Great works! thanks a lot

[Star Strider]

As always, my pleasure!