✈️ Flight Route Analysis

Machine Learning • Time Series • Route Visualization • Prediction Dashboard

This project analyzes U.S. domestic flight routes using data analytics, machine learning, and interactive geospatial visualization. It is implemented as a multi-page Streamlit dashboard covering data exploration, feature engineering, model training, forecasting, and interactive route mapping.

---

🌟 Key Features

1️⃣ Dataset Exploration (EDA)

• View dataset summary, schema, missing values.
• Visualize distributions for fare, passengers, and distance.
• Filter and export insights.

2️⃣ Feature Engineering

• Automated preprocessing (handling nulls, scaling, encoding).
• Domain-informed feature creation:
  • Seasonal indicators
  • Distance buckets
  • Demand categories
  • Fare normalization
• Reusable modular utilities via utils/.

3️⃣ Model Training & Evaluation

• Train multiple ML models:
  • Linear Regression
  • Random Forest
  • Gradient Boosting
• Compare performance using $RMSE$, $MAE$, and $R^2$.
• Visualize prediction errors and residuals.

4️⃣ Time Series Analysis

• Monthly trend analysis.
• Seasonality and year-over-year comparison.
• Time-based forecasting (ARIMA / Prophet).
• Trend decomposition plots.

[Image of a time series decomposition plot showing trend, seasonality, and residual components]

5️⃣ Prediction & Ranking

• Predict flight fares or passenger volumes.
• Rank routes by:
  • Highest predicted fare
  • Most demand
  • Cheapest predicted routes
• Interactive controls + downloadable CSV.

6️⃣ Route Visualization (Geospatial Map)

• Folium-powered USA route map.
  • Route thickness = passenger volume
  • Route color = average fare
  • City-to-city popups with route details.
• Sidebar controls:
  • Fare filter
  • Passenger filter
  • Number of routes to display
  • Opacity
  • Color scheme selection
• Dynamic plotting (Seaborn & Matplotlib):
  • Fare distribution
  • Passenger distribution
  • Fare vs. passenger scatter plot

---

📁 Project Structure

Flight_Route_Analysis/
│
├── app.py                      # Main Streamlit app entry point
├── README.md
├── requirements.txt
│
├── data/
│   └── flight_data.csv         # Dataset (~63 MB)
│
├── pages/                      # Streamlit multipage files
│   ├── 1_dataset_eda.py
│   ├── 2_feature_engineering.py
│   ├── 3_model_training.py
│   ├── 4_time_series_analysis.py
│   ├── 5_prediction_ranking.py
│   └── 6_route_visualization.py
│
└── utils/                      # Reusable functions
    ├── preprocessing.py
    ├── feature_engineering.py
    └── modeling.py

🚀 How to Run This Project

1. Install Dependencies

pip install -r requirements.txt

2. Launch the App

streamlit run app.py

The dashboard will open automatically at: http://localhost:8501

🧠 Technologies Used

• Python 3.x
• Streamlit
• Pandas / NumPy
• Scikit-learn
• Folium & streamlit-folium
• Matplotlib & Seaborn
• Statsmodels / Prophet (if used for forecasting)

📊 Dataset Information

• The dataset includes U.S. domestic flight routes with:

  • Origin & destination cities
  • Geocoded coordinates
  • Monthly passenger counts
  • Average fares
  • Route distance
  • Time period indicators
  • Large file size: ~63 MB, loaded efficiently with caching.

🔮 Future Enhancements

• Add real-time API for live airfare updates.
• Integrate LSTM or Prophet models for more accurate forecasting.
• Add clustering to identify route demand groups.
• Build airline-specific dashboards.
• Add performance benchmarking for models.

📝 License

This project is open-source under the MIT License.