CO₂ Emission Prediction using Deep Learning

Predicting CO₂ emissions using a Deep Neural Network (DNN) built with TensorFlow/Keras. This project demonstrates how deep learning can model complex relationships between multiple input features and CO₂ emissions using a large dataset.

Overview

This project trains a feedforward neural network to predict CO₂ emissions from five input features. The model is implemented using TensorFlow/Keras and trained on a dataset containing more than 7,300 samples.

The goal of this project is to compare a deep learning solution with a traditional machine learning approach and understand how neural networks perform on larger datasets.

Features

• 🧠 Deep Neural Network using TensorFlow/Keras
• 📊 Regression model for CO₂ emission prediction
• 📁 Large dataset (7,300+ samples)
• 📈 Training loss visualization
• 📉 Mean Squared Error (MSE) evaluation
• 🔍 Model architecture visualization
• ⚡ Adam optimizer
• 📚 Clean and beginner-friendly implementation

Technologies Used

• Python 3
• TensorFlow / Keras
• Pandas
• Scikit-learn
• Matplotlib
• keras-visualizer

Dataset

The dataset contains more than 7,300 samples and five input features used to predict CO₂ emissions.

Example workflow:

• Load the dataset using Pandas
• Separate features and target values
• Split the data into training and testing sets
• Train the neural network
• Evaluate prediction accuracy

Neural Network Architecture

Input Layer (5 Features)
        │
        ▼
Dense Layer (20 neurons, ReLU)
        │
        ▼
Output Layer (1 neuron)

Model Summary

| Layer | Units | Activation | | | -: | - | | Input | 5 | - | | Dense | 20 | ReLU | | Output | 1 | Linear |

Training Configuration

Parameter	Value
Optimizer	Adam
Loss Function	Mean Squared Error (MSE)
Epochs	200
Batch Size	Default
Train/Test Split	80% / 20%

Project Structure

CO2-Emission-Prediction-Deep-Learning/
│
├── co2.csv
├── co2 prediction.ipynb
├── model_DL.png
├── screenshots/
│   ├── model.png
│   └── loss.png
├── requirements.txt
├── LICENSE
└── README.md

Installation

Clone the repository:

git clone https://github.com/Matin-python/CO2-Emission-Prediction-Deep-Learning.git

Move into the project directory:

cd CO2-Emission-Prediction-Deep-Learning

Install the required packages:

pip install -r requirements.txt

or install manually:

pip install tensorflow pandas scikit-learn matplotlib keras-visualizer

How to Run

python co2 prediction.ipynb

The program will:

1. Load the dataset.
2. Build the neural network.
3. Train the model.
4. Predict CO₂ emissions.
5. Evaluate the model using Mean Squared Error.
6. Plot the training loss.
7. Save a visualization of the network architecture.

Results

The project reports:

• Mean Squared Error (MSE)
• Final training loss
• Training loss curve

These metrics help evaluate the performance of the neural network.

Screenshots

Neural Network Architecture

Training Loss

Related Projects

This project is part of a series exploring different approaches to CO₂ emission prediction.

📈 CO₂ Emission Prediction using Linear Regression

A traditional machine learning implementation using Linear Regression from scikit-learn.

Comparison

| Linear Regression | Deep Learning | | | | | Machine Learning | Deep Learning | | Linear Regression | Neural Network | | 500 samples | 7,300+ samples | | 3 input features | 5 input features | | Fast training | Longer training time | | Simple and interpretable | Can model more complex relationships |

The Linear Regression project demonstrates how a traditional regression algorithm performs on a smaller dataset, while this project shows how a neural network can learn from a larger dataset with more input features.

➡️ Repository: https://github.com/Matin-python/CO2-Emission-Prediction-Linear-Regression

Future Improvements

• 📊 Feature normalization
• 🎯 Hyperparameter tuning
• 🧠 Deeper neural network architectures
• 🔄 Cross-validation
• 📈 Additional evaluation metrics (MAE, R²)
• 💾 Save and load trained models
• ⚡ Early stopping
• 📉 TensorBoard visualization
• 🔍 Automatic hyperparameter optimization
• 🌐 Deploy the model as a web application

Contributing

Contributions, suggestions, and improvements are welcome. Feel free to fork the repository and submit a pull request.

License

This project is licensed under the MIT License.

Author

Mohammad Reza Bakhshandeh

Electrical Engineering (Electronics) Graduate

Interested in Python Development, Machine Learning, Computer Vision, Artificial Intelligence, and Data Science.