✈️ FlyPredict – Flight Fare Prediction App

🚀 A Machine Learning powered web application built using Streamlit to predict flight ticket prices and compare multiple regression models in real time.

🔗 Live App:
👉 https://flypredict.streamlit.app/

---

📌 Project Overview

FlyPredict is an end-to-end Predictive Analytics & Machine Learning project that predicts airline ticket prices using real-world features such as:

• Airline
• Source & Destination
• Departure & Arrival Time
• Number of Stops
• Travel Class
• Journey Duration
• Days Left for Booking

This project demonstrates:

• Complete ML pipeline from preprocessing to deployment
• Multi-model comparison
• Interactive visual dashboards
• A production-ready Streamlit application

---

📘 Notebook Workflow

This project follows a structured machine learning workflow:

1. Importing Libraries – Essential Python libraries imported
2. Loading the Dataset – Cleaned flight dataset loaded
3. Data Preprocessing – Handling missing values & removing unwanted columns
4. Feature Engineering – Encoding categorical features
5. Exploratory Data Analysis (EDA) – Data visualization & pattern analysis
6. Splitting Data – Train-test split
7. Model Training – Training multiple regression models
8. Model Evaluation – Performance evaluation using metrics
9. Model Comparison – Selecting the best performing model
10. Feature Importance – Identifying key influencing factors
11. Predictions – Manual input-based prediction
12. Conclusion – Final insights and recommendations

---

🧠 Machine Learning Models Used

Model	Description
✅ Random Forest Regressor	High-accuracy ensemble model
✅ K-Nearest Neighbors (KNN)	Pattern-based prediction
✅ Decision Tree Regressor	Rule-based learning
✅ Linear Regression	Baseline comparison model

✅ All models are optimized for:

• Small file size
• Fast prediction
• GitHub & Streamlit Cloud compatibility

---

✅ Key Features

✅ Real-time flight fare prediction
✅ Comparison of 4 ML models
✅ Interactive dark-themed UI
✅ Route, airline, class & timing based prediction
✅ Market insights & analytics
✅ Optimized model size for deployment
✅ Streamlit Cloud ready

---

🗂️ Project Structure

FlyPredict/
│
├── Config/
│ └── feature_order.json
│
├── Dataset/
│ └── Clean_Dataset.csv
│
├── Models/
│ ├── model_rf.pkl
│ ├── model_dt.pkl
│ ├── model_knn.pkl
│ ├── model_lr.pkl
│ ├── encoders.pkl
│ └── scaler.pkl
│
├── app.py
├── FlightFarePrediction.ipynb
├── requirements.txt
└── README.md

---

⚙️ Tech Stack

• Python
• Streamlit
• Pandas
• NumPy
• Scikit-learn
• Plotly
• Joblib

---

📊 Visualizations Included

• Airline Market Share
• Flight Class Distribution
• Price Distribution by Airline
• Booking Day vs Price Trend
• Duration vs Price Scatter Plot
• Model-wise Price Comparison

---

▶️ How to Run Locally

1️⃣ Clone the Repository

git clone https://github.com/your-username/FlyPredict.git
cd FlyPredict

2️⃣ Create Virtual Environment (Optional)

python -m venv venv
venv\Scripts\activate

3️⃣ Install Dependencies

pip install -r requirements.txt

4️⃣ Run the App

streamlit run app.py

👨‍💻 Author

Ayush Kumar 🎓 B.Tech CSE – Lovely Professional University 📊 Aspiring Data Analyst & Machine Learning Enthusiast

⭐ Support

If you like this project, consider giving it a ⭐ on GitHub!

📝 License

This project is licensed under the MIT License.