Software Project - Clustering Algorithms

Introduction

This project involves implementing a clustering algorithm using Symmetric Non-negative Matrix Factorization (SymNMF). We also built a K-means clustering algorithm for comparison on various datasets. A highlight of our project is a custom Python module, written in C, aimed at enhancing clustering performance.

Requirements

• Python 3.x
• C Compiler
• Numpy
• Pandas

📂 Structure

• symnmf.py : Python implementation of SymNMF
• symnmf.c : C implementation of SymNMF
• symnmf.h : C header file
• symnmfmodule.c : Python C API Wrapper for SymNMF
• analysis.py : Analysis and comparison script
• setup.py : Build script for Python C extension
• Makefile : Make script for C executable
• Text_file_input_example.txt Data points input file example

🚀 Quick Start

Follow these steps to get your clustering project up and running:

1️⃣ Compile C Code for SymNMF

Open your terminal and execute:

make

2️⃣ Build Python Extension

In the terminal, run:

python3 setup.py build_ext --inplace

3️⃣ Execute the Python Program

Run the SymNMF algorithm with:

python3 symnmf.py <k> <goal> <file_name.txt>
Where:

• <k>: Specifies the number of clusters
• <goal>: Defines the algorithm goal (choices are symnmf, sym, ddg, norm)
• <file_name.txt>: Data points input file (e.g., Text_file_input_example.txt)

Text File Input Format:
Such as Text_file_input_example.txt, should contain vectors separated by commas. Each line in the file represents a single data point in the feature space.

4️⃣ Conduct Analysis

To perform the data analysis, execute:
python3 analysis.py <file_name.txt>

Contributors

• Lior Kovtun
• Shalev Baruch