This repository implements a minimum distance to class mean classifier using Euclidean distances. The classifier is implemented in the classifier.m file, which calls train_classifier.m in order to train the classifier using provided training sets and then calls run_classifier.m in order to run the classifier against a test set and determine an error rate. Additionally, the samples of the training set, resulting class means, decision boundaries, and decision regions can be plotted using plot_dec_boundaries.m.
The training set can be tested on three different data sets: synthetic1.mat, synthetic2.mat, and wine.mat. The third data set was obtained from the UCI machine learning repository (http://archive.ics.uci.edu/ml/datasets/Wine). The wine.mat dataset is used to classify the cultivar of the grape plant a wine was made from, given measured attributes of the wine. Unlike the first two data sets, wine.mat contains 13 different features, so find_best_features.m can be used to narrow down the two best features to use for classification using the minimum distance to class mean classifier.
- Load data set <name.mat>:
load('data/<name.mat>') - Classify set with 2 features:
classifier(feature_train, label_train, feature_test, label_test) - Classify set with more than 2 features (use f1 and f2):
classifier(feature_train(:, [f1 f2]), label_train, feature_test(:, [f1 f2]), label_test) - Find the best features if more than 2 are available:
find_best_features(feature_train, label_train)