Plotting Classification Loss Functions

How can I plot the logisitic loss function, MathWorks says that is supports loss functions that you can specify by using the 'LossFun' name-value pair argument. The value for this function is 'logit'. So is there a way that MatLab can automically generate this function for me?

Answer 1

[Meet]

Hi Johnny,

MATLAB provides a function "fitclinear" that you can use to fit a logistic regression model. You can also specify the loss function, in your case, "logit".

Here is an example code for the same, where I have generated 100 samples for each class and labeled the data as "+1" and "1" for the two classes:

The "Learner", "logistic" argument specifies that we are using a logistic regression model.

The "LossFun", "logit" argument specifies that we are using the logistic loss function.

numSamples = 100; 

X = [randn(numSamples, 2) + 1; randn(numSamples, 2) - 1]; % Features 
y = [ones(numSamples, 1); -ones(numSamples, 1)]; % Output Binary labels: +1 and -1 

% Plot the synthetic data 
figure; 
gscatter(X(:,1), X(:,2), y, 'rb', 'xo'); 
xlabel('Feature 1'); 
ylabel('Feature 2'); 
title('Synthetic Data'); 
legend('Class +1', 'Class -1'); 
grid on; 

% Fit a Model using Logistic Loss 
mdl = fitclinear(X, y, 'Learner', 'logistic', 'LossFun', 'logit'); 
disp(mdl); 

For more information, you can refer to the following documentation link:

fitclinear: https://www.mathworks.com/help/stats/fitclinear.html