An interactive data mining application built from scratch in C++ that implements the K-Means clustering algorithm and visualizes its behavior in real-time using the SDL2 library.
I got this idea from one of my Data Analytics class where the topic were market basket analysis,fp trees and clustering
I used the two moons dataset to show the challenges of applying centroid based techniques to non-spherical data.
📊 Project Showcase: Visualizing K-Means
The primary goal is to visually demonstrate both the behavior and limitations of the K-Means algorithm.
🔹 Initial State
The visualization begins with two randomly placed centroids. All data points are initially assigned to the first cluster (red)
🔸 Final Converged State
After several iterations (by pressing the space bar), the algorithm converges. The final result clearly shows how K-Means incorrectly splits crescent-shaped clusters, highlighting a major limitation.
✅ 1. Prerequisites A C++ compiler (recommended: g++ with MinGW-w64 on Windows) SDL2 development library, placed under an SDL2/ directory
🧱 2. Compilation Open a terminal, navigate to the root directory, and run:
bash
g++ main.cpp -ISDL2/include -LSDL2/lib -lmingw32 -lSDL2main -lSDL2 -o output/visualizer.exe
📦 3. Prepare to Run Copy the SDL2 runtime file before running:
copy SDL2/bin/SDL2.dll output/
🚀 4. Execution Run the application:
./output/visualizer.exe