Analytics Portfolio Project — Production-oriented analytics platform designed to transform raw retail transaction data into actionable business insights. This project simulates a real-world analytics environment by integrating data engineering, exploratory analysis, KPI development, and dashboard reporting into a scalable workflow.

retail-analytics/
├── data/
│   ├── raw/                    # Synthetic data generated by scripts
│   └── processed/              # Cleaned, feature-engineered datasets
├── src/
│   ├── data/
│   │   ├── generate_data.py    # Synthetic data generation (Faker + NumPy)
│   │   └── preprocess.py       # ETL pipeline
│   ├── analysis/
│   │   ├── churn_analysis.py   # Problem 1: Churn scoring
│   │   ├── inventory_analysis.py # Problem 2: ABC/XYZ inventory analysis
│   │   └── forecasting.py      # Problem 3: Time-series forecasting
│   └── visualization/
│       └── chart_builder.py    # Reusable Plotly chart components
├── dashboard/
│   └── app.py                  # Plotly Dash interactive dashboard
├── notebooks/
│   └── EDA.ipynb               # Exploratory Data Analysis
├── tests/                      # Unit tests
├── .github/workflows/          # CI/CD pipeline
├── requirements.txt
└── README.md

• Python 3.11+
• macOS (M-series), Linux, or Windows

# 1. Clone the repo
git clone https://github.com/YOUR_USERNAME/retail-analytics.git
cd retail-analytics

# 2. Create virtual environment
python3 -m venv venv
source venv/bin/activate  # macOS/Linux

# 3. Install dependencies
pip install -r requirements.txt

# 4. Generate synthetic data
python src/data/generate_data.py

# 5. Run ETL pipeline
python src/data/preprocess.py

# 6. Run analysis modules
python src/analysis/churn_analysis.py
python src/analysis/inventory_analysis.py
python src/analysis/forecasting.py

# 7. Launch dashboard
python dashboard/app.py
# Open http://localhost:8050

• 23.4% of customers identified as high-risk churn (RFM score < 30)
• Top churn drivers: days since last purchase > 90, avg order value declining
• Recommended intervention: targeted email campaign for 2,847 at-risk customers

• 31% of SKUs classified as "C" items (low value, high holding cost)
• 12 products identified with chronic stockout patterns causing ~$180K lost sales
• EOQ model applied to reduce carrying costs by estimated 18%

• Prophet model achieves MAPE of 8.3% on 90-day regional forecast
• Q4 Northeast region projected at +22% YoY growth
• Southwest underperforming forecast by 14% — flagged for root cause analysis

• End-to-end analytics pipeline design
• Translation of raw data into business-relevant insights
• Application of structured analytical frameworks
• Production-aware development and deployment practices
• Clear communication of analytical results for decision-making

• Customer lifetime value modeling
• Machine learning-based churn prediction
• Forecast backtesting and model optimization
• Market basket analysis
• Real-time data pipeline integration

I, Matthew Trigg, built this project to demonstrate end-to-end analytical thinking, engineering discipline, and business communication.

MIT License — see LICENSE