silhouette-analysis

Using the Elbow Method and Silhouette Analysis to find the optimal K in K-Means Clustering.

Unsupervised ML analysis of lifestyle data to uncover risk patterns for Dry Eye Disease

Analysis to optimize services & resident satisfaction in senior living facilities by segmenting population based on characteristics & behaviors.

A clustering evaluation framework that combines micro- and macro-averaged silhouette scores into a composite metric using statistical weighting.

Confidence in Assignments via K-partition Ensembles

Unsupervised machine learning

Unsupervised K-Means clustering of 198 BBMP wards into four climate-vulnerability archetypes, derived from satellite measurements of surface temperature, urbanisation, vegetation, and water-body proximity. Independently corroborates the cool-roof prioritisation produced by the companion supervised UHI model.

This project explores customer segmentation and market analysis in the context of online retail using an online retail dataset. By applying advanced analytics, we aim to uncover insights that can drive strategic decisions and enhance business performance.

Unsupervised Learning - Using K Means algorithm to Cluster the customers.

Unsupervised ML project segmenting 660 AllLife Bank credit card customers using K-Means (k=3, silhouette=0.52) and Hierarchical Clustering (cophenetic=0.90). Includes EDA, clustering validation (elbow, silhouette, cophenetic), and business-actionable cluster profiling to guide targeted marketing and service improvements.

A modular, research-grade Python library for unsupervised learning with embeddings (PCA, t-SNE, UMAP) and clustering (KMeans, DBSCAN, GMM). Includes reproducible experiments, metrics, visualizations, and tests—perfect for ML research and coursework.

Data Mining - EDA, Feature Selection, Standardize, Remove Global Outliers, Normalize, Feature Extraction (with PCA), Clustering, Classification (baseline models and hyperparameter tuning with GridSearchCV).

Customer clustering using silhouette K-means and silhouette analysis on Python. Also using logistic regression on Python to predict top 30 customers.

The project uses KMeans clustering on the Global Superstore dataset to categorize customers based on their buying habits, aiming to help retailers make better business decisions by tailoring their marketing strategies and improving their inventory management.

CompSil: Composite Silhouette for Cluster-Count Selection

Exact and approximate silhouette scoring with micro, macro, and cluster-weighted averages.

A machine learning project that segments retail customers using K-Means and Hierarchical Clustering. The project includes data preprocessing, feature engineering, PCA, cluster evaluation, and visualization to generate actionable customer insights for targeted marketing.

Description: "Data preprocessing, K-Means clustering with Silhouette analysis, and PCA dimensionality reduction on microclimate, obesity, and gene expression datasets. Achieved 267% clustering improvement with PCA."

K-means clustering of chess games: feature engineering, optimal-K search, silhouette analysis, PCA visualization, and interpretable cluster archetypes. Full 10-step professional project plan included.

Creating predictive models to classify Trump's vote share and clustering counties based on demographics and economic variables. Report findings in PDF with detailed methodologies, model assessments, and R code for the project.