🛒 Retail Vendor Performance Analysis

📌 Project Summary

This project delivers an end-to-end data analysis pipeline to evaluate vendor performance in a retail environment. It focuses on identifying profitability drivers, optimizing procurement decisions, and improving inventory efficiency using data-driven insights.

---

🎯 Business Problem

Retail companies often struggle with:

• Lack of visibility into vendor-wise profitability
• Inefficient procurement strategies leading to increased costs
• Overstocking or understocking due to poor inventory planning
• Over-reliance on a small subset of vendors

This project addresses these issues by analyzing vendor transactions, inventory flow, and pricing data.

---

🚀 Key Objectives

• Analyze vendor-wise sales, purchases, and profit contribution
• Identify top-performing and underperforming vendors
• Evaluate inventory turnover and stock efficiency
• Detect pricing inefficiencies and cost leakages
• Provide actionable recommendations for business optimization

---

🧠 Approach & Methodology

1. Data Ingestion & Cleaning

• Imported multiple datasets (sales, purchases, inventory)
• Handled missing values and removed duplicates
• Standardized column formats for consistency

2. Data Transformation

• Aggregated vendor-level sales and purchase data
• Merged datasets to create a unified analytical view
• Engineered new features such as:
  • Profit = Sales − Purchase Cost
  • Profit Margin (%)
  • Inventory Turnover Ratio

3. Analysis

• Vendor-wise performance evaluation
• Profitability and margin analysis
• Inventory efficiency assessment
• Identification of high-risk and high-opportunity vendors

4. Visualization

• Bar charts for top vendors by profit
• Distribution plots for profit margins
• Interactive dashboard using Power BI

---

🛠️ Tech Stack

• Python (Pandas, NumPy) → Data processing & analysis
• Matplotlib → Visualization
• SQL (optional) → Querying structured data
• Power BI → Interactive dashboard
• Jupyter Notebook → Exploratory Data Analysis
• Git & GitHub → Version control

---

📂 Project Structure

Retail-Vendor-Performance-Analysis/
│
├── data/ # Raw datasets
├── notebooks/ # EDA notebooks
├── scripts/ # Modular Python scripts
│ ├── data_processing.py
│ ├── analysis.py
│ └── visualization.py
├── outputs/ # Processed outputs
├── main.py # Pipeline entry point
├── requirements.txt
└── README.md

---

📊 Key Metrics Computed

• Total Sales per Vendor
• Total Purchase Cost
• Profit (Revenue − Cost)
• Profit Margin (%)
• Inventory Turnover Ratio

---

📈 Key Insights

• A small group of vendors contributes disproportionately to total revenue → vendor dependency risk
• Several vendors show high margins but low sales → growth opportunity
• Bulk purchasing significantly reduces per-unit cost → procurement optimization
• Low inventory turnover indicates capital inefficiency and overstocking

---

💡 Business Impact

• Enables data-driven vendor selection
• Reduces procurement costs through optimized purchasing
• Improves inventory planning and reduces wastage
• Supports strategic decision-making with measurable insights

---

⚙️ How to Run

1. Clone the Repository

git clone https://github.com/nirjanadas/Retail-Vendor-Performance-Analysis.git
cd Retail-Vendor-Performance-Analysis

2. Install Dependencies

pip install -r requirements.txt

3. Execute the Pipeline

python main.py

4. View Outputs

• Processed results stored in /outputs
• Visualizations displayed during execution
• Open Power BI dashboard (.pbix) for interactive insights

---

🧪 Sample Output

• Top vendors ranked by profit
• Profit margin distribution
• Inventory turnover insights

---

🔮 Future Enhancements

• Build predictive model for vendor risk scoring
• Automate ETL pipeline using Airflow
• Deploy dashboard to Power BI Service / Web
• Integrate real-time data streams

---

🧠 Skills Demonstrated

• Data Cleaning & Preprocessing
• Feature Engineering
• Exploratory Data Analysis (EDA)
• Data Visualization
• Business Analytics & Insight Generation Modular Code Design

---

👤 Author

Nirjana Das

GitHub:nirjanadas