Power BI pages
6
Executive, sales, returns, segments, risk, quality
End-to-End E-commerce Analytics & ML Decision System
RetailPulse
A complete portfolio simulation that generates recurring e-commerce data, validates quality issues, loads clean data into PostgreSQL, builds SQL analytics views, exports dashboard-ready tables, trains ML models, and produces Power BI, Python, and business-reporting outputs.
Current version: v0.7.0
Data generation, validation, SQL analytics, dashboard exports, modeling, reporting, Power BI assets, Python visuals, linting, and tests are complete.
SQL views
6
Sales, product, RFM, shipping, returns, churn features
Dashboard exports
15+
CSV outputs for BI and reporting
Best forecast MAPE
10.39%
Linear regression validation result
Business Context
From raw monthly operations data to management-ready outputs.
RetailPulse answers recurring monthly questions an e-commerce analytics team would support: revenue and profit trends, product and category performance, data quality risks, revenue forecasting reliability, customer inactivity risk, and management actions for the latest month.
Workflow
1Generate synthetic base tables
2Generate monthly transaction batches
3Inject controlled data quality issues
4Validate raw monthly files
5Load clean data to PostgreSQL
6Create SQL analytical views
7Export dashboard tables
8Run forecasting, segmentation, and inactivity models
9Produce Power BI, Python visuals, and a monthly business review
Business Goal
Questions the analytics system is designed to answer.
How are revenue, profit, orders, returns, and delivery performance trending?
Which categories, products, and customer groups drive commercial outcomes?
Which data quality issues appear before reporting and dashboard delivery?
Which customers are likely to become inactive and need retention action?
What should management review in the latest monthly business cycle?
Business Impact
How the outputs support business decisions.
Executive monitoring through KPI cards, category performance, and trend views.
Planning support through revenue forecasting and model comparison.
Retention support through customer segmentation and inactivity risk scoring.
Operational improvement through returns, late deliveries, and data quality checks.
BI readiness through clean SQL views and dashboard-ready export tables.
Business Question Map
How each dashboard module connects to a decision.
| Business Question | Dashboard Module | Decision Use |
|---|---|---|
| Which categories are profitable? | Category Profitability | Prioritize margin, growth, and return-reduction actions. |
| How are KPIs changing over time? | Monthly KPI Trend | Separate growth signals from operational risk. |
| Which forecasting model is most reliable? | Forecasting Lab | Choose a model that is accurate and explainable. |
| Which customers may become inactive? | Inactivity Risk Explorer | Build targeted retention and win-back audiences. |
| Can the reporting data be trusted? | Data Quality Monitor | Identify table-level issues before dashboard delivery. |
SQL Analytics Layer
- vw_monthly_sales
- vw_product_profitability
- vw_customer_rfm
- vw_shipping_performance
- vw_return_analysis
- vw_churn_features
Business Outputs
- KPI scorecards and executive monitoring visuals
- Power BI dashboard pages built from cleaned CSV exports
- Automated monthly business review in Markdown
- Revenue forecasting with model comparison
- Customer segmentation and recommended actions
- Customer inactivity prediction and risk scoring
What It Demonstrates
- Reproducible analytics workflow
- Data validation before reporting
- BI-ready export design
- Machine learning with business interpretation
- Clear project communication
- Tests and task orchestration
Project Repository
Review the complete code, pipeline, tests, and original project assets on GitHub.
The portfolio page focuses on an interactive analysis experience. The full RetailPulse repository remains the source for the implementation, notebooks, Power BI file, generated reports, and exported data.
Open RetailPulse Repo