Analytics & Forecasting Platform

Production-ready platform delivering end-to-end e-commerce analytics and forecasting—featuring interactive KPI dashboards, ML-driven A/B test analysis, customer segmentation, CLV modeling, and demand forecasting to drive data-driven growth

🔗 Live Demo • 📜 License • 📂 Source

Installation

1. Create & Activate a Virtual Environment

# macOS / Linux
python3 -m venv venv
source venv/bin/activate

# Windows
python -m venv venv
.\venv\Scripts\Activate.ps1

2. Install Dependencies

pip install -r requirements.txt

3. Run the App Locally

streamlit run app.py

Architecture

1. Extract & Ingest: daily CSV files → Pandas

2. Transform & Clean: missing values, outlier filtering, date normalization

3. Analytics & Modeling

   • A/B testing with Statsmodels
   • RFM → PCA → K-Means segmentation
   • CLV prediction (Random Forest)
   • Demand forecasting (Prophet)

4. Serve & Visualize: Streamlit front-end with Plotly

Data Dictionary

Column	Type	Description
order_id	string	Unique order identifier
order_date	datetime	Purchase timestamp
customer_id	string	Unique customer identifier
revenue	float	Order value in USD
segment	string	Customer RFM cluster label

Key Features & Modules

KPI Dashboard

• Total Revenue: $800,323
• Avg. Order Value (AOV): $80.03
• Total Orders: 10,000
• Interactive time-series & boxplots by category/payment

A/B Testing

• Single-run Z-test → +4.5% lift (p < 0.05)
• 100 bootstrapped simulations → 95% CI [2.1%, 6.9%]

from statsmodels.stats.proportion import proportions_ztest
zstat, pval = proportions_ztest([success_A, success_B], [n_A, n_B])

Customer Segmentation (ML)

• RFM → PCA → K-Means (k=4)
• Interactive cluster scatter

CLV Modeling (ML)

• RandomForestRegressor (MAE 85.3 | RMSE 111.7)

from sklearn.ensemble import RandomForestRegressor
model = RandomForestRegressor(random_state=42).fit(X_train, y_train)

Sales Forecasting

• Prophet with trend/seasonality decomposition
• MAPE: 9.8% on hold-out data

from prophet import Prophet
m = Prophet(); m.fit(df.rename(columns={'date':'ds','sales':'y'}))

Key Insights

1. Premium Segment drives +20% AOV → target for premium bundles
2. A/B Test shows +4.5% conversion lift → rollout advised
3. Forecast Accuracy (MAPE 9.8%) → enables 12% stockout reduction

Business Impact

• Inventory Planning: 12% fewer stockouts via accurate forecasts
• Marketing ROI: +15% incremental revenue from targeted clusters
• Analytics Efficiency: 70% faster A/B test analysis vs. manual

Roadmap

• Real-time streaming & anomaly detection
• Automated A/B test recommendations
• CI/CD deployment on AWS/GCP
• BI integrations (Power BI, Tableau)

Tech Stack

• Python 3.9+ (pandas, NumPy, statsmodels)
• ML: scikit-learn, Prophet
• Visualization: Streamlit, Plotly
• Containerization: Docker
• Badges & Graphics: Shields.io

License

Distributed under the MIT License.