Gaussian mixture models, k-means, mini-batch-kmeans and k-medoids clustering

Image Segmentation using Superpixels, Affinity Propagation and Kmeans Clustering

Jax implementation of Mini-batch K-Means algorithm

This project consists in the implementation of the K-Means and Mini-Batch K-Means clustering algorithms. This is not to be considered as the final and most efficient algorithm implementation as the objective here is to make a clear comparison between the sequential and parallel execution of the clustering steps.

Color compression of an image with K-Means Clustering Algorithm which can help in devices with low processing power and memory for large images

Developed for "Management and Analysis of Physics Dataset Mod. B," this project uses Dask and CloudVeneto VMs to handle a massive 250GB dataset. Clustering on 800k RCV1 articles involves dataset reduction by macrocategory and also implementing cosine similarity for improved clustering, as suggested by Natural Language Processing principles.

Performing basic clustering on a seeds dataset.

This project used a Kmeans after PCA model to segment retail customers to optimize marketing efforts. When the model repeatedly returned a single cluster, the model was used to prove the customers' homogenous characteristics. Influenced the bank's marketing strategies and initiatives. Developed in Jupyter Notebook with Python for FNB.