✈️ Flight Delay Classification Using Machine Learning

This project explores the classification of flight delays using classic machine learning models, aiming to predict whether a flight will be delayed based on various factors such as departure time, airline, and airport data.

📌 Project Overview

• Developed as part of final project assessment for ELCE 455: Machine Learning with Python (Fall 2024) course with instructor Professor Dr. Amin Zollanvari
• Dataset: US domestic flights (January 2019)
• Best performing model: Random Forest (92% accuracy, 0.77 F1-score for delayed flights)
• Implemented models: Logistic Regression, Decision Trees, Random Forest, KNN, AdaBoost

📊 Methodology

1. Data Preprocessing: Handling missing values, encoding categorical variables, and feature scaling.
2. Feature Engineering: Adding temporal features and spatial-temporal interactions for improved predictive power.
3. Model Training & Selection: Evaluated multiple ML models using cross-validation and grid search.
4. Evaluation Metrics: Accuracy, Precision, Recall, F1-score, and ROC-AUC.

🚀 Results & Insights

• Random Forest outperformed other models, effectively handling non-linear feature interactions.
• AdaBoost performed well but had slightly lower recall for delayed flights.
• Feature engineering, including departure delay indicators and time-based binning, significantly improved model performance.
• Future improvements: Incorporating weather data and deep learning models like LSTMs.

🔧 Installation & Usage

📦 Requirements

• Python 3.9+
• Required libraries:

  pip install pandas numpy scikit-learn matplotlib seaborn

▶️ Running the Model

1. Clone the repository:

   git clone https://github.com/tvran/Forte-stt.git
   cd Forte-stt

2. Run the preprocessing and model training:

   python main.py

📂 Project Structure

📁 Forte-stt/
│── Predicting_the_Delay_of_Flights_Turan.ipynb                # Project notebook
│── README.md                                                  # Project documentation
│── dictionaries.py                                            # Loaded dataset

📌 Institution: Nazarbayev University, School of Engineering and Digital Sciences, Department of Electrical and Computer Engineering
📅 Date: November 24, 2024