📈 Retail Sales Prediction (Big Mart Sales)

A Machine Learning regression project that predicts Item_Outlet_Sales for retail products based on product and outlet characteristics, using the Big Mart Sales dataset.

🌐 Live Demo

👉 👉 Try it live here

🧠 How it works

1. Load and clean the dataset (fill missing Item_Weight with mean, Outlet_Size with mode)
2. Standardize inconsistent labels in Item_Fat_Content (LF, low fat, Low Fat → Low Fat)
3. One-hot encode categorical features
4. Train and compare Linear Regression and Random Forest Regressor
5. Select the best-performing model and use it for predictions

📊 Model Performance

Model	R² Score	MAE	RMSE
Linear Regression	unstable (severe overfitting from raw ID columns)	—	—
Random Forest ✅	~0.56	~765	~1094

🛠️ Tech Stack

• Python, Pandas, NumPy
• scikit-learn (RandomForestRegressor, LinearRegression)
• Matplotlib (visualizations in the notebook)
• Streamlit (web app)

📂 Files

• Sales_Prediction_Machine_Learning.ipynb — full analysis notebook (EDA, visualizations, model comparison)
• app.py — interactive Streamlit web app
• Train.csv — training dataset
• testdata.csv — test dataset
• requirements.txt — Python dependencies

▶️ How to run locally

Notebook version:

pip install -r requirements.txt
jupyter notebook Sales_Prediction_Machine_Learning.ipynb

Web app version:

pip install -r requirements.txt
streamlit run app.py

📌 Future Improvements

• Hyperparameter tuning (GridSearchCV) for Random Forest
• Try Gradient Boosting / XGBoost for potentially better accuracy
• Add cross-validation for more robust performance estimates