app.py

import streamlit as st
import pandas as pd
import numpy as np
import pickle
import matplotlib.pyplot as plt

# Load model
with open('random_forest_classifier (1).pkl', 'rb') as file:
    model = pickle.load(file)

# Define the possible outcomes
class_labels = ['phishing', 'benign', 'defacement', 'malware']

def main():
    st.set_page_config(page_title="Malicious URL Detection", layout="wide")
    
    st.title("Malicious URL Detection")
    
    st.markdown("""
        <style>
        .main {background-color: #f5f5f5;}
        </style>
        """, unsafe_allow_html=True)
    
    st.markdown("### Enter the URL to predict its category")
    
    url_input = st.text_input("Enter URL")
    
    if st.button("Predict"):
        if url_input:
            # Perform prediction
            prediction, prediction_probabilities = predict_url(url_input)
            display_prediction(prediction, prediction_probabilities)
        else:
            st.warning("Please enter a URL for prediction.")

def predict_url(url):
    # Logic to extract features from the URL and perform prediction
    # For demonstration purposes, let's just return a random prediction and random probabilities
    prediction = np.random.choice(class_labels)
    prediction_probabilities = np.random.rand(len(class_labels))
    prediction_probabilities /= np.sum(prediction_probabilities)  # Normalize to get probabilities
    return prediction, prediction_probabilities

def display_prediction(prediction, prediction_probabilities):
    st.subheader(f"The URL is predicted to be: {prediction}")
    
    if prediction == 'phishing':
        st.error("The URL is predicted to be: phishing")
    elif prediction == 'malware':
        st.warning("The URL is predicted to be: malware")
    elif prediction == 'defacement':
        st.warning("The URL is predicted to be: defacement")
    else:
        st.success("The URL is predicted to be: benign")
    
    # Display prediction probabilities
    st.subheader("Prediction Probabilities:")
    prob_df = pd.DataFrame(prediction_probabilities, index=class_labels, columns=["Probability"])
    st.bar_chart(prob_df)

    
    # Another example chart: Prediction distribution (simulated)
    st.subheader("Prediction Distribution (Simulated):")
    prediction_counts = np.random.randint(1, 100, size=len(class_labels))
    prediction_counts[class_labels.index(prediction)] += 100  # Simulate the current prediction
    
    fig, ax = plt.subplots()
    ax.bar(class_labels, prediction_counts, color=['red', 'green', 'blue', 'orange'])
    ax.set_xlabel('Class')
    ax.set_ylabel('Count')
    ax.set_title('Prediction Distribution')
    st.pyplot(fig)

if __name__ == '__main__':
    main()