Evaluate the model's performance with some parameters like
- Accuracy
- Precision
- Recall
- F1 score
and finally plot the figures of some useful parameters.
% -------------------------------------------------------
% [t,Accuracy,Precision,Recall,F1score] = evaluate(x1,x2,n)
% -------------------------------------------------------
% inputs:
% x1 -- 1-D array with arbitrary length (positive data)
% x2 -- 1-D array with arbitrary length (negative data)
% n -- Divide x1 and x2 into n pieces
% -------------------------------------------------------
% outputs:
% t -- X-axis of threshold
% Accuracy -- Accuracy of positive data
% Precision -- Precision of positive data
% Recall -- Recall of positive data
% F1score -- F1score of positive data
% -------------------------------------------------------
>> npieces = 100
>> positive_data = randn(5e4,1)*2+3.0;
>> negative_data = randn(6e4,1)*3-5.0;
>> evaluate(positive_data, negative_data, npieces);
Just make sure that the positive data is on the right side of negative data.