Performance_Evaluation.m

clear; close all; clc

%% Assessing clustering methods' performance and efficacy
% Cluster analysis organizes data into groups based on similarities between
% the data points. Sometimes the data contains natural divisions that 
% indicate the appropriate number of clusters. Other times, the data does 
% not contain natural divisions, or the natural divisions are unknown. In 
% such a case, you might to determine the optimal number of clusters to 
% group your data. (from MATLAB)

%% Part 1- Creating a fake dataset
nRow_fake_dataset = 3000; % it should actually be 30000, but for elapsed 
% running time of the code, 3000 is chosen.  
nCol_fake_dataset = 600; % it should actually be 6000, but for elapsed 
% running time of the code, 600 is chosen.

temp0 = randn(ceil(nRow_fake_dataset/5), ceil(nCol_fake_dataset/5));
temp1 = repmat(temp0, 5, 5);
fake_dataset = temp1(1:nRow_fake_dataset, 1:nCol_fake_dataset);

%% Part 2- Performing linear dimensionality reduction (or noise suppression) alg.
[coeff, score, ~, ~, explained, ~] = pca(fake_dataset);
explained_variance_threshold = 99.5; % for the FFT dataset, a threshold of 
% 99 or 99.5 is thought to be more suitable.

cumulative_explained = cumsum(explained);
numFeatures = find(cumulative_explained >= explained_variance_threshold);
numFeatures = numFeatures(1);
reduced_dataset_by_pca = score(:, 1:numFeatures);
inputs = reduced_dataset_by_pca.';

%% Part 3- Investigating different clustering methods (1-SOFM & 2-LVQ)
numClusters = 3;

%% Part 3-1-1- Training SOFM
row_nodes = 3; column_nodes = 1;
% Note: the result of the product "row_nodes * column_nodes", should be
% equal to "numClusters".

dimensions = [row_nodes, column_nodes];
SOFM_net = selforgmap(dimensions);
SOFM_net = train(SOFM_net, inputs);
SOFM_oneHotClasses = SOFM_net(inputs); SOFM_classes = vec2ind(SOFM_oneHotClasses).';

%% Part 3-1-2- Evaluating SOFM
SOFM_CalinskiHarabasz_index = evalclusters(inputs.', SOFM_classes, ...
    'CalinskiHarabasz')
SOFM_DaviesBouldin_index = evalclusters(inputs.', SOFM_classes, ...
    'DaviesBouldin')
SOFM_silhouette_index = evalclusters(inputs.', SOFM_classes, ...
    'silhouette')

% Creating a silhouette plot from the clustered data.
figure;
silhouette(inputs.', SOFM_classes); title(['Silhouette Plot for ', '"SOFM"'])
grid on

%% Part 3-2-1- Training Competitive Layer
numEpochs = 100;
LVQ_net = competlayer(numClusters);
configure(LVQ_net, inputs);
LVQ_net.trainParam.epochs = numEpochs;

LVQ_net = train(LVQ_net, inputs);
LVQ_oneHotClasses = LVQ_net(inputs); LVQ_classes = vec2ind(LVQ_oneHotClasses).';

%% Part 3-2-2- Evaluating Competitive Layer
LVQ_CalinskiHarabasz_index = evalclusters(inputs.', LVQ_classes, ...
    'CalinskiHarabasz')
LVQ_DaviesBouldin_index = evalclusters(inputs.', LVQ_classes, ...
    'DaviesBouldin')
LVQ_silhouette_index = evalclusters(inputs.', LVQ_classes, ...
    'silhouette')

% Create a silhouette plot from the clustered data.
silhouette(inputs.', LVQ_classes); title(['Silhouette Plot for ', '"LVQ"'])
grid on