BRSbl4z3/AI_Based_Network_Intrusion_Detection_System

1import streamlit as st2import pandas as pd3import numpy as np4from sklearn.ensemble import RandomForestClassifier5from sklearn.model_selection import train_test_split6from sklearn.metrics import accuracy_score7from groq import Groq8import os9 10# --- PAGE SETUP ---11st.set_page_config(page_title="AI-NIDS Student Project", layout="wide")12 13st.title("AI-Based Network Intrusion Detection System")14st.markdown("""15**Student Project**: This system uses **Random Forest** to detect Network attacks and **Groq AI** to explain the packets.16""")17 18# --- CONFIGURATION ---19DATA_FILE = "Friday-WorkingHours-Afternoon-DDos.pcap_ISCX.csv"20 21# --- SIDEBAR: SETTINGS ---22st.sidebar.header("1. Settings")23groq_api_key = st.sidebar.text_input("Groq API Key (starts with gsk_)", type="password")24st.sidebar.caption("[Get a free key here](https://console.groq.com/keys)")25 26st.sidebar.header("2. Model Training")27 28@st.cache_data29def load_data(filepath):30    try:31        df = pd.read_csv(filepath, nrows=15000)32        df.columns = df.columns.str.strip()33        df.replace([np.inf, -np.inf], np.nan, inplace=True)34        df.dropna(inplace=True)35        return df36    except FileNotFoundError:37        return None38 39def train_model(df):40    features = ['Flow Duration', 'Total Fwd Packets', 'Total Backward Packets', 41                'Total Length of Fwd Packets', 'Fwd Packet Length Max', 42                'Flow IAT Mean', 'Flow IAT Std', 'Flow Packets/s']43    target = 'Label'44    45    missing_cols = [c for c in features if c not in df.columns]46    if missing_cols:47        st.error(f"Missing columns in CSV: {missing_cols}")48        return None, 0, [], None, None49 50    X = df[features]51    y = df[target]52    53    X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=42)54    55    clf = RandomForestClassifier(n_estimators=10, max_depth=10, random_state=42)56    clf.fit(X_train, y_train)57    58    score = accuracy_score(y_test, clf.predict(X_test))59    return clf, score, features, X_test, y_test60 61# --- APP LOGIC ---62df = load_data(DATA_FILE)63 64if df is None:65    st.error(f"Error: File '{DATA_FILE}' not found. Please upload it to the Files tab.")66    st.stop()67 68st.sidebar.success(f"Dataset Loaded: {len(df)} rows")69 70if st.sidebar.button("Train Model Now"):71    with st.spinner("Training model..."):72        clf, accuracy, feature_names, X_test, y_test = train_model(df)73        if clf:74            st.session_state['model'] = clf75            st.session_state['features'] = feature_names76            st.session_state['X_test'] = X_test 77            st.session_state['y_test'] = y_test78            st.sidebar.success(f"Training Complete! Accuracy: {accuracy:.2%}")79 80st.header("3. Threat Analysis Dashboard")81 82if 'model' in st.session_state:83    col1, col2 = st.columns(2)84    85    with col1:86        st.subheader("Simulation")87        st.info("Pick a random packet from the test data to simulate live traffic.")88        89        if st.button("🎲 Capture Random Packet"):90            random_idx = np.random.randint(0, len(st.session_state['X_test']))91            packet_data = st.session_state['X_test'].iloc[random_idx]92            actual_label = st.session_state['y_test'].iloc[random_idx]93            94            st.session_state['current_packet'] = packet_data95            st.session_state['actual_label'] = actual_label96            97    if 'current_packet' in st.session_state:98        packet = st.session_state['current_packet']99        100        with col1:101            st.write("**Packet Header Info:**")102            st.dataframe(packet, use_container_width=True)103 104        with col2:105            st.subheader("AI Detection Result")106            prediction = st.session_state['model'].predict([packet])[0]107            108            if prediction == "BENIGN":109                st.success(f" STATUS: **SAFE (BENIGN)**")110            else:111                st.error(f"🚨 STATUS: **ATTACK DETECTED ({prediction})**")112            113            st.caption(f"Ground Truth Label: {st.session_state['actual_label']}")114 115            st.markdown("---")116            st.subheader(" Ask AI Analyst (Groq)")117            118            if st.button("Generate Explanation"):119                if not groq_api_key:120                    st.warning(" Please enter your Groq API Key in the sidebar first.")121                else:122                    try:123                        client = Groq(api_key=groq_api_key)124                        125                        prompt = f"""126                        You are a cybersecurity analyst. 127                        A network packet was detected as: {prediction}.128                        129                        Packet Technical Details:130                        {packet.to_string()}131                        132                        Please explain:133                        1. Why these specific values (like Flow Duration or Packet Length) might indicate {prediction}.134                        2. If it is BENIGN, explain why it looks normal.135                        3. Keep the answer short and simple for a student.136                        """137 138                        with st.spinner("Groq is analyzing the packet..."):139                            completion = client.chat.completions.create(140                                model="llama-3.3-70b-versatile",  # <--- UPDATED MODEL NAME141                                messages=[142                                    {"role": "user", "content": prompt}143                                ],144                                temperature=0.6,145                            )146                            st.info(completion.choices[0].message.content)147                            148                    except Exception as e:149                        st.error(f"API Error: {e}")150else:151    st.info(" Waiting for model training. Click **'Train Model Now'** in the sidebar.")