Customer Segmentation App
ML-powered CRM analytics for actionable business intelligence.
Project Overview
Businesses often sit on massive datasets without clear segmentation. This app bridges the gap by providing a self-service tool for RFM (Recency, Frequency, Monetary) analysis, allowing non-technical stakeholders to identify high-value customer segments through data-driven clustering.
Key Features
Automated RFM Calculation
Intelligent mapping of transactional data into Recency, Frequency, and Monetary scores.
Interactive KMeans Clustering
Dynamic cluster sizing with visual feedback using Scikit-learn.
Dynamic Data Export
Generate summaries and export segmented customer lists for targeted marketing.
System Architecture
A Python-centric data processing pipeline focused on speed and mathematical accuracy.
Processing Layer
Data cleaning, encoding detection, and feature engineering for RFM scores.
ML Engine
KMeans implementation with dynamic hyperparameter tuning for optimal clustering.
Visualization
Interactive 3D and 2D charts for cluster distribution and density analysis.
Engineering Challenges
Handling inconsistent date formats and monetary currency symbols across various user-uploaded datasets.
Developed a robust regex-based preprocessing engine that automatically sanitizes and normalizes datetime and float columns.
Ensuring high-speed clustering for datasets exceeding 100k rows in a browser environment.
Implemented efficient data vectorization with NumPy and optimized the KMeans initialization state (k-means++).
Screenshot Gallery
Tech Stack
My Role
- Architected the data pipeline and ML model selection.
- Implemented the Streamlit frontend for interactive parameter tuning.
- Created the automated RFM calculation logic and visualization suite.