Harmony search based clustering algorithm

Harmony search based clustering on synthetic data set

This set of files perform Harmony search based clustering algorithm.
The proposed novel partitional clustering approach extracts information in the form of optimal cluster centers from training samples. The extracted cluster centers are then validated on test samples.
This illustration contains two files, namely, Main_fn.m and Harmony_Search.m
Main_fn.m is the main file which generates synthetic data. Post the training phase, clustering is carried out on test data set and results are displayed

Harmony_Search.m contains optimal cluster center extraction from the training dataset using Harmony search. The file takes in training dataset, upper & lower limits of data and number of attributes as the input. The file returns the optimal cluster center to the Main_fn.m.

In the main file, a synthetic data is generated with predefined mean and standard deviation. The users can vary these parameters. The users can also implement algorithm using there own datasets. The dataset and corresponding training, testing portion should replace the variables xdata, ftrain, ftest with section of lines 20-45 in main_fn.m

The result of clustering can be visualized through confusion matrix in line 122-124 and Overall Accuracy in lines 130-133 of Main_fn.m

Note: Please do not switch between figure windows during program execution.