EricoR/Indian_Airplane_Price_Prediction

1import streamlit as st2import pandas as pd3import numpy as np4import pickle5import json6 7def run():8    st.title('Flight Data Prediction')9 10    # Project Summary Section11    st.write('### Summary of Flight Analysis')12    st.write(13        """14        This flight data analysis covers flight volume, popular routes, and the factors that influence ticket prices.15 16        **Flight Volume and Popular Routes:**17        * **Vistara** is the airline with the highest number of flights, followed by **Air India** and **Indigo**.18        * **Delhi** and **Mumbai** are the main flight hubs, with the highest volume of flights to and from other major cities like **Bangalore**.19        * The highest number of departures occurs in the morning.20        """21    )22    st.markdown('---')23 24    st.write('### Factors Influencing Ticket Prices')25    st.write(26        """27        Flight ticket prices are not static and are influenced by several key factors:28        * **Airline Type:** Airlines can be grouped into two pricing categories:29            * **Low-Cost Carriers:** Such as **AirAsia**, **Indigo**, and **GO_FIRST**, have relatively stable and affordable prices.30            * **Full-Service Carriers:** Such as **Vistara** and **Air India**, have significantly higher prices and a more varied price range.31        * **Departure and Arrival Times:** Ticket prices vary depending on the combination of departure and arrival times.32        * **Distance Traveled:** Routes with longer distances, which require more fuel, tend to have higher ticket prices.33        * **Time of Booking:** The closer the purchase date is to the departure date, the more ticket prices tend to increase.34        """35    )36    st.markdown('---')37 38    st.write('### Model Selection and Performance')39    st.write(40        """41        To predict ticket prices, a variety of regression models were evaluated. Simple linear models like Linear Regression and Ridge Regression performed poorly due to the data's complex, non-linear nature. Advanced tree-based models, particularly **boosting models** like **XGBoost** and **LightGBM**, were far more effective at capturing these complex patterns.42 43        After comparing the models, **XGBoost** was selected as the best choice. Hyperparameter tuning was performed on the model to optimize its performance, resulting in the following final metrics:44        * **Mean R² Score: 0.9787**45          This indicates that the model can explain approximately **98%** of the variance in ticket prices, signifying a very high level of accuracy.46        * **Mean Negative MSE: -10,948,750**47          This score shows that the model has a very low average prediction error, making it a reliable tool for forecasting airline ticket prices.48        """49    )50    st.markdown('---')51    52    # Load model53    with open('src/ModelXGB.pkl', 'rb') as file_1:54        model = pickle.load(file_1)55    st.success("Model loaded successfully!")56 57    st.write('## Input Data')58    with st.form(key='data'):59        airlines = ['SpiceJet', 'AirAsia', 'Vistara', 'IndiGo', 'Akasa Air']60        flight_numbers = [f'SG-{np.random.randint(1000, 9999)}' for _ in range(5)] + \61                         [f'I5-{np.random.randint(100, 999)}' for _ in range(5)] + \62                         [f'UK-{np.random.randint(900, 999)}' for _ in range(5)] + \63                         [f'6E-{np.random.randint(100, 999)}' for _ in range(5)] + \64                         [f'QP-{np.random.randint(100, 999)}' for _ in range(5)]65        source_cities = ['Delhi', 'Mumbai', 'Bangalore', 'Kolkata', 'Chennai']66        destination_cities = ['Mumbai', 'Delhi', 'Bangalore', 'Kolkata', 'Chennai']67        departure_times = ['Early_Morning', 'Morning', 'Afternoon', 'Evening', 'Night']68        arrival_times = ['Early_Morning', 'Morning', 'Afternoon', 'Evening', 'Night']69        stops = ['zero', 'one', 'two_or_more']70        classes = ['Economy', 'Business']71        72        airline = st.selectbox('Airline', airlines)73        flight_number = st.selectbox('Flight Number', flight_numbers)74        source_city = st.selectbox('Source City', source_cities)75        destination_city = st.selectbox('Destination City', destination_cities)76        departure_time = st.selectbox('Departure Time', departure_times)77        arrival_time = st.selectbox('Arrival Time', arrival_times)78        stops = st.selectbox('Stops', stops)79        flight_class = st.selectbox('Class', classes)80        duration = st.number_input('Duration (hours)', min_value=1.5, max_value=4.0, step=0.1)81        days_left = st.number_input('Days Left for Departure', min_value=1, max_value=60)82 83        # Submit button84        submit = st.form_submit_button('Predict')85 86    if submit:87        data = {88            'airline': airline,89            'flight': flight_number,90            'source_city': source_city,91            'destination_city': destination_city,92            'departure_time': departure_time,93            'arrival_time': arrival_time,94            'stops': stops,95            'class': flight_class,96            'duration': duration,97            'days_left': days_left,98        }99        data = pd.DataFrame([data])100        st.dataframe(data)101 102        # Make prediction103        prediction = model.predict(data)104        st.write(f'Prediction: {prediction[0]}')105 106# Call the run function to execute the Streamlit app107if __name__ == "__main__":108    run()