Mainabryan/SMART-_ECOMMERCE

ShopSmart E-commerce Analytics + GA4 + Ads (ML & BI) Solving poor marketing ROI, high churn, and unclear sales insights for an African e-commerce brand. Using Python (ML), SQL, and Tableau to predict churn, forecast CLV, optimize ad spend, and visualize multi-country performance across Kenya, Nigeria, South Africa, Ghana, and Egypt. ShopSmart E-commerce Analytics + GA4 + Ads (ML & BI) This end-to-end data science and business intelligence project focuses on solving key e-commerce challenges such as poor marketing ROI, rising customer churn, and limited visibility into regional sales performance. The project is built for ShopSmart, a multi-country African e-commerce platform operating in Kenya, Nigeria, South Africa, Ghana, and Egypt. It demonstrates how data-driven insights can directly improve decision-making and profitability across multiple markets. # Project Objective To integrate analytics from GA4, ads data, and e-commerce transactions to understand customer behavior, forecast lifetime value, detect anomalies, and optimize marketing investments. This project unifies raw data into actionable insights through Python, SQL, and Tableau workflows. # Key Problems Solved Identify and reduce customer churn through predictive modeling Forecast Customer Lifetime Value (CLV) for better marketing segmentation Optimize ad spending and measure cross-channel ROI Detect fraudulent or bot-driven activity in orders Improve sales forecasting and regional performance tracking # Machine Learning (Python – Google Colab) Churn Prediction (Random Forest) – ROC AUC: 0.88 CLV Forecasting (Gradient Boosting) – MAE: $42 Product Recommendation System (SVD) – Personalized Top-3 products Ad ROI Optimization (Random Forest) – TikTok ROAS: 4.2x Anomaly Detection (Isolation Forest) – Flags suspicious orders Each model includes feature engineering, model evaluation, and business interpretation. # Business Intelligence (Tableau) Interactive dashboards provide: Revenue and AOV trends Country-level sales mapping Churn and CLV segmentation Ad funnel performance ML-driven VIP and risk alerts # Tech Stack Python (Pandas, scikit-learn, Surprise), Google Colab, Tableau Public, SQL-ready CSVs. # Deliverables Clean datasets ML notebooks and results Tableau workbook (.twb) Two-page insights report (PDF) …