plot_1feature_2clas​s

Visualises usefulness of a feature for binary classification.

When you attempt to classify data in 2 classes based on a single feature, the first thing to do is to review the conditional probability of the point being from the first class given the value of the feature. However, as usual, overfit is creeping in the shadows. Therefore, the amount of data that the conditional probability is based on should be reviewed as well. This function plots on the same graph the conditional probability and histogram of data distribution for each class separately.
For example, let's assume that you are attempting to predict whether the patient will show paroxysmal atrial fibrillation during 24 hour ambulatory ECG recording (stored in boolean vector 'holter') based on value of peak ischemic ST depression during cardiac stress test (stored in vector 'ST').
Use the following commands to initialize the variables:
>> N = 1e3;
>> k = 20;
>> ST = [5+randn(1,N) 2*rand(1,k) 8+2*rand(1,k)];
>> holter = [2+6*rand(1,N)<ST(1:N) true(1,k-1) false(1,k) true];

Now you can visualise the data using the following commands:
>> plot_1feature_2class(ST, holter, 'ST, mm', {'PAF', 'no PAF'})

For additional explanations type: help plot_1feature_2class

This submission is courtesy of Norav Medical (www.norav.com) - the leading company in the fields of PC-ECG, EKG Management systems and related non-invasive cardiac devices.