How can I fit a function using fmincon?

Question

0 个投票

have data from a movement control experiment in which participants were timed while they performed a task, and given a score (failure or success) based on their performance. As you might expect, we assume participants will make less errors as they have more time to perform the task.

Visualizing the data gives me:

I can fit my desired model to this using cftoolbox, but a least squares fit doesn't make sense for binary data. Just to illustrate:

My attempts to fit the model using fmincon are below. This returns the error "Error using fmincon (line 609) Supplied objective function must return a scalar value."

Could anybody help me change my code to get a fit to the following sample data?

time = [12.16 11.81 12.32 11.87 12.37 12.51 12.63 12.09 11.25
7.73 8.18 9.49 10.29 8.88 9.46 10.12 9.76 9.99 10.08
7.48 7.88 7.81 6.7 7.68 8.05 8.23 7.84 8.52 7.7 
6.26 6.12 6.19 6.49 6.25 6.51 6 6.79 5.89 5.93 3.97 4.91 4.78 4.43
3.82 4.72 4.72 4.31 4.81 4.32 3.62 3.71 4.29 3.46 3.9 3.73 4.15
3.92 3.8 3.4 3.7 2.91 2.84 2.7 2.83 2.46 3.19 3.44 2.67 3.49 2.71
3.17 2.97 2.76 2.71 2.88 2.52 2.86 2.83 2.64 2.02 2.37 2.38
2.53 3.03 2.61 2.59 2.59 2.44 2.73 ]
error = [0  0   0   0   0   0   0   0   0   1   0   0   0   0   1   1   1
 0   0   0   0   1   1   1   1   1   1   0   0   0   0   1   1   1   0   1
 0   1   0   1   1   0   1   1   0   1   1   1   1   1   1   1   1   1   1 
1   1   0   1   1   1   1   1   1   1   1   1   1   1   1   1   1   1
 1   1   1   1   1   1   1   1   1   1   1   1   1   1   1   1];
%Code:
% initial parameters - a corresponds to params(1), b corresponds to params(2)
a = 3.0;
b = -0.01;
LL = @(params) 1/1+params(1)*(log(time).^params(2));
LL([a b]); 
pOpt = fmincon(LL,[a b],[],[]);