README.md

Clustering Algorithm Implementation and Visualization from Scratch with Python

Overview

This project implements four popular clustering algorithms from scratch in Python, designed to work for datasets with d >= 2 dimensions and k >= 2 clusters. The implementations are tested on 2D datasets and compared visually with scikit-learn's implementations to evaluate correctness and performance.

Implemented Clustering Algorithms

1. K-Means Clustering
2. Gaussian Mixture Model (GMM) using Expectation-Maximization (EM)
3. Mean-Shift Clustering
4. Agglomerative Clustering

---

Python Implementations

• KMeans.py: K-Means clustering.
• KMeans_Ver0.py: K-Means clustering (2nd version).
• GaussianMM.py: EM-GMM.
• GaussianMM_Ver0.py: EM-GMM with functions of AIC, BIC and predict (2nd version).
• MeanShift.py: Mean-Shift clustering.
• Agglomerative.py: Agglomerative clustering.

---

Evaluations and Tests

• test_2d_visualization.py:
  Tests each implementation on 2D datasets with visualization, comparing the results to scikit-learn's equivalent algorithms.
• data_2d_test/:
  Contains the datasets used for testing.
• test_2d_visualization_results/:
  Stores the output images of the clustering results.

---

Visualization Results

Blobs Dataset

Algorithm	My Implementation	Scikit-learn
Agglomerative
EM-GMM
K-Means
Mean-Shift

---

Moons and Stars Dataset

Algorithm	My Implementation	Scikit-learn
Agglomerative
EM-GMM
K-Means
Mean-Shift

---

Sticks Dataset

Algorithm	My Implementation	Scikit-learn
Agglomerative
EM-GMM
K-Means
Mean-Shift