A complete end-to-end data generation, cleaning, and visualization project comparing two business years (2023 vs 2024) through custom synthetic data. Built entirely in Python (Jupyter Notebook) and visualized using Power BI — combining storytelling with interactivity using dynamic slicers, maps, and custom tooltips.

Python (pandas, Faker, NumPy) – Generated and cleaned 2023 & 2024 datasets using random logic and libraries.

Power BI – Designed interactive dashboards with city/state maps, slicers, donut charts, KPI cards, tooltips, and bar graphs.

Generative AI (ChatGPT) – Used as a co-pilot to debug scripts, organize logic, rewrite markdown, and brainstorm layout/design enhancements.

GitHub: Link

LinkedIn: Link

Portfolio: Link

This project doesn’t use real-world business sales data. Instead:

🔹 I first generated synthetic sales data for 2023 and 2024 using Python’s Faker and NumPy libraries. 🔹 Then saved them into two datasets:

While most people look at just total revenue per year, this project dives deeper into:

🛒 Are certain product categories more dominant by sales or by total orders?

🏙️ Which cities contributed the most to overall revenue?

📈 How do trends change month-by-month across KPIs like Sales, Orders, Quantity Sold?

👤 Who are the top customers and how much are they contributing to our growth?

🧭 What is the business growth direction comparing 2023 and 2024?

These questions often get ignored in surface-level reports — this project is about answering them clearly and interactively.

Used Faker, random, and NumPy to simulate customer/product/order behaviour.

Created two different logic blocks — one for 2023 and another for 2024 (with product-category mappings, customer/product ID formatting, etc.)

Ensured consistent column names, formats, and realistic date ranges.

Final output was saved as .csv files to be imported into Power BI.

➡️ Notebook: SalesDataset.ipynb

This dashboard combines a clean layout with multiple pages to show different stories across product, region, and customer KPIs.

📄 Page 1: KPI Dashboard

📌 KPI cards: Actual sales, quantity sold, orders, customers

📈 Line charts: Month-wise sales and orders

🎯 Gauge charts: Year-end vs target metrics (sales, orders, customers)

📄 Page 2: Comparison Analysis

📊 Total sales by product category (bar chart)

🧮 Total orders by product category (donut)

🗺️ Total sales by city (map view)

🧑 Top 100 customers (ranked by sales & quantity sold)

📄 Page 3: Custom Tooltip View

This view appears as a tooltip when users hover over visuals (e.g., sales chart).

Designed as a Qtr-wise breakdown of all KPIs.

📈 2024 saw a significant increase in both total customers and total sales.

📚 Categories like Clothing and Electronics were the most ordered and highest selling.

🗺️ Cities like San Francisco, New York, and Houston led the sales map.

💎 Top 100 customers accounted for over 60% of the total sales.

🎯 Almost every metric beat its original yearly target.

How to create realistic synthetic datasets with controlled randomness.

Clean structuring of logic blocks across multiple simulated datasets.

Setting up well-formatted data to be dashboard-ready.

Designing Power BI dashboards that are interactive, not overwhelming.

Leveraging Generative AI to write cleaner markdowns, handle exceptions, and make quick adjustments in logic.

📥 Download both sales datasets from the repo folder (Dataset/)

🐍 Run the Python notebook to understand the data generation logic.

📊 Open the Power BI .pbix file and explore dashboards/slicers/tooltips.

🧭 Use custom slicers to filter per category, city, or customer and watch visuals change dynamically.

➡️ Power BI File: Sales_Data_Dashboard.pbix 🖼️ Published version: [Add Power BI service link here]