Amazon Sales Analytics Dashboard
A comprehensive analytics project on Amazon sales transaction data. Covers the full analytics lifecycle: SQL-based data preparation, exploratory data analysis in Python, machine learning forecasting models, and a 4-page Tableau dashboard deployed for stakeholder use.
Problem
Raw Amazon sales data is transactional — every row is an order. Without aggregation, modeling, and visualization, it's impossible to identify which products drive revenue, where seasonal patterns exist, or how demand varies by region and category.
Solution
SQL transformations cleaned and aggregated raw transactional records. Python EDA identified key patterns. Scikit-learn models produced demand forecasts. A 4-page Tableau dashboard (Welcome, Executive Summary, Revenue Analysis, Item Analysis) delivered actionable insights to business stakeholders.
Architecture
SQL CTEs and window functions cleaned, deduplicated, and aggregated raw order records into analytics-ready tables by product, region, and time period.
Python (Pandas, Matplotlib, Seaborn) explored distributions, seasonal trends, top SKUs, and regional patterns. Key findings informed dashboard design.
Scikit-learn regression models trained on historical order data to forecast demand by product category and region. Cross-validated with RMSE and MAE metrics.
4-page Tableau workbook: Welcome (KPI summary), Executive (high-level trends), Revenue Analysis (time-series breakdowns), Item Analysis (SKU-level performance).
Highlights
- 4-page Tableau dashboard covering executive, revenue, and item-level views.
- SQL window functions and CTEs for clean dimensional aggregation.
- ML demand forecasting models with cross-validation and error metrics.
- EDA surfaced seasonal patterns and top-performing SKUs by region.