🔍 Corporate Sales Data Warehouse: Exploratory Data Analysis (EDA) using SQL

Deep-Dive Dimensional Analytics & Business Intelligence on the Gold Layer

⚠️ Data Dependency Note: The analytical database queried in this repository is dynamically populated and hosted by the pipeline built in the Corporate Sales Data Warehouse (SQL Server) Project. It directly interrogates the production-ready Gold Layer star schema.

An advanced, script-driven Exploratory Data Analysis (EDA) project executed natively via Transact-SQL (T-SQL) on Microsoft SQL Server. This project serves as the analytical audit layer directly built upon the verified Gold Layer (Business Ready) dimensional star schema from the Corporate Sales Data Warehouse ecosystem.

🏗️ Methodological Framework

The analysis follows a disciplined, 6-stage structured framework designed to sweep through systemic metadata down to isolated transactional outliers.

---

📝 Detailed Analytical Phases & SQL Script Strategy

🗄️ 1. Database Exploration (Metadata Auditing)

• Objective: Map out the systemic landscape of the Gold consumption layer, catalog objects, and analyze volume footprints.
• T-SQL Tactics Used: Programmatic interrogation of sys.objects, sys.columns, and sys.partitions metadata tables to dynamic data schema tracking.
• Business Insight: Verified table structure alignments and calculated exact physical row counts across the Star Schema constraints to establish baseline data volumes before running mathematical operations.

👥 2. Dimensions Exploration (Categorical Cardinality Analysis)

• Objective: Isolate and analyze the categorical attributes that define our business realities (dim_customers, dim_products).
• T-SQL Tactics Used: DISTINCT aggregations, multi-field concatenation audits, and string tracking queries.
• Business Insight: Discovered the total unique customer footprint, monitored product catalog skewness, tracked distribution attributes, and ensured that dimensional attributes contain no unmapped or orphaned keys.

🔄 3. Data Exploration (Data Profiling & Completeness)

• Objective: Audit the data quality and structural health of the fact records, analyzing temporal ranges and completeness constraints.
• T-SQL Tactics Used: Non-null boundary checking (COUNT(column) vs COUNT(*)), chronological boundaries (MIN(), MAX()), and conditional filter profiles (WHERE).
• Business Insight: Established the definitive historical window of our sales ledger, validated relational data alignment across the Date dimension, and mathematically verified that zero revenue leakages or orphaned transactions exist within the schema boundaries.

📊 4. Measures Exploration (Statistical Metric Aggregations)

• Objective: Compute foundational statistical summaries across all core metrics to map out statistical tendencies and structural boundaries.
• T-SQL Tactics Used: SUM(), AVG(), MIN(), MAX(), STDEV(), and mathematical division ratios.
• Business Insight: Generated exact, baseline corporate figures for critical metrics like Average Revenue Per User (ARPU), aggregate gross revenue volumes, and structural profit/tax ratios across global transaction variables.

📈 5. Magnitude Exploration (Volumetric Outlier Tracking)

• Objective: Segment and bucket numerical quantities to observe volume distributions and identify structural anomalies or enterprise-scale transactions.

• T-SQL Tactics Used: Conditional formatting bins (CASE WHEN), discrete arithmetic grouping, and cumulative filters (HAVING).

• Business Insight: Successfully categorized transactional volumes into clear operational brackets (e.g., Low, Medium, High Value Transactions), isolating massive bulk corporate purchases from micro-consumer purchases to understand revenue distribution models.

🏆 6. Ranking Exploration (Window Function Performance Intelligence)

• Objective: Execute high-performance analytic reporting to reveal top-performing products, seasonal operational high points, and regional sales leaders.
• T-SQL Tactics Used: Advanced non-aggregating Window Functions (ROW_NUMBER(), RANK(), DENSE_RANK()) combined with partition strategies (PARTITION BY ... OVER()).
• Business Insight: Identified the top 10 generating products across individual categories and ranked store locations based on localized moving revenue trends. This provides immediate, actionable visibility into what drives the company's bottom line.

---

🛠️ Execution & Technical Implementations

• Script Modularity: All exploratory components are written in a structured, clean format with comprehensive documentation tags explaining the business intent behind each query block.
• Performance Tuning: Queries use native T-SQL window functions and group aggregations designed to take advantage of underlying primary key indexes, maximizing processing speed across large records.

---

👤 About Me

Hi, I'm Bande Abdoul-Rahmane 👋

🛠️ Data Engineer, MLOps Practitioner & AI Explorer

I specialize in building end-to-end data ecosystems—from high-performance relational architectures using the Medallion framework to containerized machine learning pipelines running in production. I design, build, and optimize every layer that data touches.

• Data Engineering: Designing structured DWH platforms (Bronze → Silver → Gold) using SQL Server and complex dimensional Star Schemas.
• Analytics Engineering: Writing advanced T-SQL engines, analytical views, and deep-dive exploratory scripts.
• MLOps Pipeline: Packaging predictive models (XGBoost, Scikit-Learn) into production environments using Docker and automated batch scoring.
• Next Horizon: Currently diving deep into Agentic AI and LLM orchestration to build self-correcting data workflows.

“Algorithms are only as good as the infrastructure supporting them.”

---

🚀 Moving to Next Phase: Advanced Analytics Engineering

With the initial exploratory profile complete, the next phase is packaging advanced calculations into permanent database layers.

👉 Proceed to: Advanced Analytics & View Engineering Project

---