Overview

Designed and built an unsupervised machine learning pipeline to cluster countries into meaningful development categories using KMeans and PCA. The project leverages Python, Pandas, NumPy, and Scikit-learn to analyze multi-dimensional socio-economic indicators and provide data-driven insights that support economic analysis, policy research, and interactive dashboard reporting.

Responsibilities

• Designed the end-to-end data preprocessing and feature scaling pipeline using StandardScaler
• Implemented KMeans clustering and applied the Elbow Method to identify the optimal number of clusters
• Evaluated model performance using Silhouette Score to assess cluster quality and separation
• Developed PCA-based visualizations to interpret high-dimensional data in a 2D analytical view
• Performed cluster profiling and labeling based on GDP, income, life expectancy, and child mortality metrics
• Integrated outputs for dashboard-ready export and country-level lookup analysis

Outcome

• Enabled clear segmentation of countries into development categories for faster economic analysis
• Improved interpretability of complex, high-dimensional data through visual clustering and profiling
• Supported better data-driven decision-making for reporting, research, and storytelling use cases

Conclusion:

This project strengthened my skills in unsupervised learning, data interpretation, and analytical visualization, and can be easily scaled into interactive BI dashboards or extended with additional global indicators for deeper insights.