Lufemage Data Lab: End-to-End Sales Analysis

This project simulates a complete data analysis workflow for a junior data analyst role. It demonstrates the ability to handle a data project from inception to final reporting, including data generation, cleaning, analysis, insight generation, and visualization.

🚀 Interactive Dashboard

This project has been deployed as an interactive web application using Streamlit. You can explore the data and filter the results in real-time.

➡️ View the Interactive Dashboard Live

Project Overview

As a Junior Data Analyst at Lufemage Labs, I was tasked with creating a functional prototype for sales analysis. The objective was to showcase the lab's capability to transform raw data into actionable business intelligence. Since no pre-existing data was provided, this project includes a custom script to generate a realistic synthetic dataset simulating sales transactions.

---

Project Phases

The analysis followed a structured, multi-phase approach:

1. Phase 1: Synthetic Data Generation: A Python script using Faker and NumPy was developed to create a realistic dataset of 5,000 sales records, including customer IDs, purchase dates, amounts, product categories, cities, and payment methods.
2. Phase 2: Exploratory Data Analysis (EDA): The dataset was loaded into a Jupyter Notebook for initial exploration. This included data cleaning (e.g., converting data types), checking for duplicates/nulls, and calculating descriptive statistics.
3. Phase 3: Insight Generation: Deep-dive analysis was performed using pandas to answer key business questions by grouping and aggregating data.
4. Phase 4: Data Storytelling: The findings were synthesized into a coherent narrative, transforming raw analysis into a compelling story for stakeholders.
5. Phase 5: Data Visualization: Key insights were visualized using Matplotlib and Seaborn to create clear, impactful charts and a final summary dashboard.

---

Key Business Questions Addressed

• Which products or categories generate the most revenue?
• Which cities are the top-performing markets?
• Are there specific times (day of the week, hour) when sales peak?
• What are the most popular payment methods among customers?
• What is the average transaction value (ATV) per city?

---

Tech Stack

• Language: Python 3.9
• Libraries:
  • Data Manipulation: Pandas, NumPy
  • Data Generation: Faker
  • Data Visualization: Matplotlib, Seaborn
  • Development Environment: JupyterLab

---

Project Structure

lufemage-data-lab/
├── .venv/
├── data/
│ ├── raw_sales_data.csv # Generated raw data
│ └── figures/
│    └── sales_analysis_dashboard.png # Exported dashboard image
├── notebooks/
│ └── 01_sales_analysis.ipynb # Main analysis notebook
├── scripts/
│ └── generate_dataset.py # Script to generate synthetic data
├── .gitignore
├── README.md
└── requirements.txt

---

Setup and Installation

To run this project locally, follow these steps:

1. Clone the repository:

   git clone https://github.com/Elimge/lufemage-data-lab.git
   cd lufemage-data-lab

2. Create and activate a virtual environment:

   # For macOS/Linux
   python3 -m venv .venv
   source .venv/bin/activate
   
   # For Windows
   python -m venv .venv
   .venv\Scripts\activate

3. Install the required dependencies:

   pip install -r requirements.txt

---

How to Run the Project

1. Generate the dataset: Run the generation script from the root directory. This will create raw_sales_data.csv inside the data/ folder.

   python scripts/generate_dataset.py

2. Run the analysis notebook: Start JupyterLab and open the notebook located in the notebooks/ directory.

   jupyter lab

   Navigate to notebooks/01_sales_analysis.ipynb and run the cells to see the complete analysis.

---

Key Findings & Dashboard

The analysis revealed a clear profile of the primary customer: an urban, tech-savvy consumer, likely from Bogotá, who plans purchases for the weekend and prefers digital payment methods.

Summary Dashboard

Actionable Recommendations

1. Focus on Core Strengths: Launch a hyper-segmented marketing campaign for the 'Electronics' category, targeting the Bogotá market as a testbed for new product offerings.
2. Optimize Marketing Timing: Align digital marketing campaigns to activate on Thursday afternoons and intensify during peak purchasing hours (2 PM - 6 PM) on weekends.
3. Enhance User Experience: Prioritize a technical audit of the credit card payment flow to ensure it is seamless, as it accounts for 50% of all transactions.

Executive Presentation

For a summary of the key findings and strategic recommendations, you can view the full executive presentation prepared for stakeholders.

➡️ View the Full Presentation (PDF)

Business Intelligence (BI) Dashboard Prototype

To demonstrate the integration between Python-based analysis and industry-standard BI tools, a prototype dashboard was created in Power BI. This visualizes one of the core findings: revenue distribution by product category. This step proves the data is ready for business-level reporting and monitoring.

---

Future Improvements

• Apply statistical tests to validate hypotheses (e.g., A/B testing promotional strategies).
• Build a simple predictive model to forecast sales for the next quarter.

---

Author

• Miguel Canedo Vanegas
• GitHub: @Elimge
• Email: elimge@outlook.com