This project applies unsupervised clustering techniques to a digit dataset in order to group similar data points without using labels.

• Hierarchical Clustering (Ward linkage)
• BIRCH Clustering
• Gaussian Mixture Model (GMM)

• Python
• Pandas, NumPy
• Scikit-learn
• Matplotlib

• Data was normalized before clustering
• Hierarchical and BIRCH clustering produced the best results
• Experimental approaches (proximity matrix, deep learning) were not effective

Open and run the notebook:

jupyter notebook clustering.ipynb