ZainabBee24/afrimarket-risk-analysis

A data-driven risk analysis of AfriMarket sellers using the Dataverse Africa July 2024 challenge dataset. This project covers data cleaning, seller behavior analytics, predictive modeling for return risks, and strategic recommendations for improving customer trust and delivery efficiency in African e-commerce. # 📦 AfriMarket Risk Analysis – Dataverse Africa Challenge This repository contains the full analysis and solution for the **Dataverse Africa July 2024 Challenge**: **"Jumia Jitter – Understanding Seller Behavior & Risk in AfriMarket"** --- ## 📁 Project Structure - `data/`: Cleaned dataset used for modeling and visualizations - `notebooks/`: Final analysis notebook (feature engineering, EDA, modeling, strategy) - `visuals/`: Canva and dashboard visuals used in the report - `presentation/`: Final presentation slide (risk framework, strategy, summary) - `README.md`: This documentation file ## 🧩 Tasks Covered - ✅ Data cleaning & feature engineering - ✅ Suspicious review detection - ✅ Exploratory analytics & heatmaps - ✅ Class imbalance handling (SMOTE) - ✅ Predictive models: Logistic Regression, Random Forest, XGBoost - ✅ Seller Risk Scoring & Strategy - ✅ Dashboard visualization (Power BI/Canva) - ✅ Customer Trust Policy - ## 📸 Social Media Post *Participated and tagged @DataverseAfrica on LinkedIn/Twitter.* ## 📊 Dashboard & Strategy Slides - 📍 **Dashboard Highlights**: Seller risk, complaints by region/category, prediction flags - 🛡️ **Risk Framework**: 1-slide Canva visual used in board strategy recommendation ## 🧠 Tools Used - Python (Pandas, Scikit-learn, XGBoost, Matplotlib) - Google Colab - Power BI / Canva (Dashboard & Slides) - GitHub (for submission) ## ✅ Author **Zainab Balarabe Adam** LinkedIn | adamzainabb1@gmail.com ## 📎 Submission Notes ## 🚨 Note on Streamlit Hosting Due to environment limitations and time constraints, the Streamlit app could not be deployed before the challenge deadline. However, the `app.py` file is included in this repository and can be run locally using: ```bash pip install -r requirements.txt streamlit run app.py All required assets are organized and screenshots attached per instructions. Thank you Dataverse Africa team for this impactful challenge!