AdGrass – Machine Learning Ad Recommendation Engine

Overview

AdGrass is a Machine Learning-powered advertisement recommendation engine designed to simulate personalized ad delivery during live sports events. The system predicts the most relevant advertisement category for each user using demographic information, viewing behavior, engagement history, and device type.

The project leverages a Random Forest Classifier to analyze user data and generate accurate advertisement recommendations, demonstrating how machine learning can improve user engagement and advertising efficiency.

---

Features

• Personalized advertisement recommendations
• Random Forest-based classification model
• User behavior analysis
• Demographic-based targeting
• Device-aware recommendation system
• Web interface built with Flask
• Analytics dashboard for recommendation insights
• Model persistence using Joblib
• CSV-based dataset support

---

Project Architecture

                User Data
                      │
      ┌───────────────┼───────────────┐
      │               │               │
 Demographics   Viewing History   Device Type
                      │
                Data Preprocessing
                      │
              Feature Engineering
                      │
          Random Forest Classifier
                      │
         Advertisement Recommendation
                      │
          Flask Web Application
                      │
             Recommendation Output

---

Technology Stack

Category	Technologies
Language	Python 3
Machine Learning	Scikit-learn
Data Processing	Pandas, NumPy
Backend	Flask
Model Storage	Joblib
Visualization	Streamlit (Analytics Dashboard)
Version Control	Git & GitHub

---

Machine Learning Workflow

1. Load user dataset
2. Clean and preprocess data
3. Encode categorical features
4. Split dataset into training and testing sets
5. Train Random Forest Classifier
6. Evaluate model performance
7. Save trained model
8. Deploy through Flask
9. Generate advertisement recommendations

---

Input Features

The model considers several user attributes including:

• Age
• Gender
• Location
• Preferred Sport
• Viewing Time
• Device Type
• Previous Advertisement Engagement
• Watch Duration
• Click History

---

Output

The model predicts the most suitable advertisement category, such as:

• Sports Equipment
• Electronics
• Food & Beverages
• Fashion
• Travel
• Gaming
• Health & Fitness
• Streaming Services

---

Dataset

The project uses a structured CSV dataset containing user behavioral and demographic information.

Example:

Age	Gender	Sport	Device	Watch Time	Previous Clicks	Recommended Ad
24	Male	Cricket	Mobile	120 min	Yes	Sports Equipment
31	Female	Football	Smart TV	90 min	No	Food & Beverage

---

Installation

Clone the repository:

git clone https://github.com/yourusername/AdGrass.git

Move into the project directory:

cd AdGrass

Create a virtual environment:

Windows

python -m venv .venv

Activate it:

.venv\Scripts\activate

Install dependencies:

pip install -r requirements.txt

---

Running the Project

Train the model:

python train_model.py

Run the Flask application:

python app.py

Launch the analytics dashboard:

streamlit run analytics.py

---

Project Structure

AdGrass/
│
├── app.py
├── train_model.py
├── analytics.py
├── requirements.txt
├── README.md
│
├── models/
│   └── random_forest_model.pkl
│
├── data/
│   ├── dataset.csv
│   └── ad_logs.csv
│
├── templates/
│   └── index.html
│
├── static/
│   ├── css/
│   └── images/
│
└── utils/
    └── preprocessing.py

---

Model

Algorithm

Random Forest Classifier

Why Random Forest?

• Handles categorical and numerical features effectively
• Reduces overfitting through ensemble learning
• High prediction accuracy
• Robust to noisy data
• Provides feature importance analysis

---

Evaluation Metrics

Typical evaluation metrics include:

• Accuracy
• Precision
• Recall
• F1 Score
• Confusion Matrix

Example:

Accuracy : 92.4%

Precision : 91%

Recall : 93%

F1 Score : 92%

---

Future Improvements

• Deep Learning recommendation models
• Real-time recommendation engine
• Collaborative filtering
• Hybrid recommendation system
• User feedback learning
• Cloud deployment
• REST API support
• Live sports streaming integration
• A/B testing framework

---

Applications

• Sports streaming platforms
• OTT services
• Online advertising
• Digital marketing
• Personalized recommendation systems
• Smart TV advertising
• Mobile advertising platforms

---

Learning Outcomes

This project demonstrates:

• Machine Learning model development
• Classification using Random Forest
• Data preprocessing
• Feature engineering
• Model deployment with Flask
• Analytics dashboard development
• Recommendation system design
• End-to-end ML pipeline implementation

---

Contributing

Contributions are welcome.

1. Fork the repository
2. Create a new branch

git checkout -b feature-name

3. Commit changes

git commit -m "Add new feature"

4. Push the branch

git push origin feature-name

5. Open a Pull Request

---

License

This project is licensed under the MIT License.

---

Author

Developed by: [Your Name]

Machine Learning | Artificial Intelligence | Python | Flask | Scikit-learn

---

Acknowledgements

• Scikit-learn
• Flask
• Pandas
• NumPy
• Streamlit
• Open Source Community