ProcureFlow – Procurement Analytics (Sievo-style)

Problem

Global procurement teams struggle to unify multi-source spend data (CSV/API/ERP), normalize currencies, and produce supplier insights like budget variance, regional mix, and monthly trends. The question: “Can we deliver a reliable, explainable pipeline that transforms raw spend into business-ready analytics?”

My Approach

1) Python ETL (Clean & Enrich)

• Ingested suppliers, categories, FX rates, budgets, and transactions.
• Validated keys, deduplicated Tx, converted to EUR, clipped outliers per category.
• Produced a star schema export: FactSpend, DimSupplier, DimCategory, and a flattened spend_cleaned for quick analysis.

2) SQL Star Schema

Fact-Dimension design for reliable analytics & DAX measures: FactSpend.Supplier_ID → DimSupplier.Supplier_ID, FactSpend.Category_ID → DimCategory.Category_ID.

3) Power BI Dashboard

• Spend Overview (KPI cards, category mix, regional split)
• Supplier Performance (budget variance, top suppliers)
• Trends (monthly spend + category mix, slicers)
• Data Model Overview (documentation of star schema)

Live Dashboard

If the report doesn’t load, open it directly in a new tab: Open Power BI.

Impact

• Trustworthy analytics: normalized spend in EUR, clean keys, and documented lineage.
• Business value: supplier budget variance, top-N suppliers, regional/category mix.
• Scalable design: Python ETL + SQL schema, ready to swap SQLite → Azure SQL DB.

Tech Stack

Python (pandas) • SQL (star schema) • SQLite (demo warehouse) • Power BI Desktop/Service • Azure-ready architecture

Architecture

DimSupplier ───► FactSpend ◄─── DimCategory
    

Project Files

All cleaned data files, ETL scripts, SQL queries, and documentation are available on GitHub.