{Click on image for broad clear view}

🔍 Project Overview

This project focuses on predicting sales for Big Mart stores across India using historical data. I built a machine learning model that analyzes factors like product price, store type, location, and promotions to forecast future sales with greater accuracy.

---

🎯 Business Problem

Big Mart needed a reliable way to forecast sales to:

• Avoid overstocking and stockouts

• Optimize inventory and staffing

• Plan promotions more effectively

Using machine learning helps make smarter, data-driven decisions.

---

🛠 Tools & Techniques Used

• Python + Google Colab

• Libraries: Pandas, Seaborn, Matplotlib, XGBoost, Scikit-learn

• Model Used: XGBoost Regressor

• Data Source: Kaggle – Big Mart Sales Dataset

---

⚙️ Key Steps I Followed

1. Data cleaning and handling missing values

2. Feature encoding (Label Encoding for categories)

3. Exploratory Data Analysis using Seaborn

4. Model training with XGBoost

5. Evaluation using R² Score (Train: 0.87, Test: 0.51)

---

📊 Key Findings

• Higher MRP items tend to have more sales

• Larger supermarkets in Tier 1 cities perform better

• Product visibility alone doesn't drive sales

• Holidays and promotions create strong seasonal effects

📈  Visuals findings

---

🤖 ML Model Performance

• XGBoost Regressor showed strong predictive power on training data

• Test R² score: ~0.51 — good baseline with room for feature improvements

---

For code and more information

Click Link : 📄 View Full Project Report (PDF)

---

🧠 Final Thoughts

This project helped me understand how machine learning models can be used in retail to improve decision-making. I’d love to extend this by:

• Adding external factors (like weather or holidays)

• Deploying the model into a dashboard

• Exploring deep learning-based forecasting