Update app.py

app.py

@@ -1,90 +1,51 @@
 import streamlit as st
 import pandas as pd
-import pickle
+import numpy as np
 import joblib
+import pickle
 
-# Load model artifacts
-@st.cache_resource
-def load_model_artifacts():
-    with open('features_to_drop.pkl', 'rb') as f:
-        features_to_drop = pickle.load(f)
-    with open('category_encodings.pkl', 'rb') as f:
-        category_encodings = pickle.load(f)
-    model = joblib.load('xgb_model.pkl')
-    return features_to_drop, category_encodings, model
-
-# Preprocessing function
-def preprocess_input(df, features_to_drop, category_encodings):
-    df = df.copy()
-    drop_cols = list(features_to_drop.intersection(df.columns))
-    df.drop(columns=drop_cols, inplace=True, errors='ignore')
-    for col, cats in category_encodings.items():
-        if col in df.columns:
-            df[col] = df[col].astype(str)
-            df[col] = pd.Categorical(df[col], categories=cats)
-            df[col] = df[col].cat.codes
-    df = df.fillna(0)
-    return df
-
-# UI layout
-st.set_page_config("IDS - Selected Features Input", layout="wide")
-st.title("Intrusion Detection System (IDS) - Manual Test Input")
-
-st.markdown("Enter selected features for a single row of network flow data:")
-
-with st.form("selected_input_form"):
-    col1, col2 = st.columns(2)
-    with col1:
-        dur = st.number_input("Duration of Flow (`dur`)", min_value=0.0, step=0.001)
-        proto = st.selectbox("Protocol (`proto`)", ["tcp", "udp", "icmp", "-"])
-        service = st.text_input("Service (`service`)", value="-")
-        state = st.selectbox("State (`state`)", ["CON", "INT", "FIN", "REQ", "-"])
-        sbytes = st.number_input("Source Bytes (`sbytes`)", min_value=0)
-        spkts = st.number_input("Source Packets (`spkts`)", min_value=0)
-        sttl = st.number_input("Source TTL (`sttl`)", min_value=0)
-        sload = st.number_input("Source Load (`sload`)", min_value=0.0)
-
-    with col2:
-        dbytes = st.number_input("Destination Bytes (`dbytes`)", min_value=0)
-        dpkts = st.number_input("Destination Packets (`dpkts`)", min_value=0)
-        sloss = st.number_input("Source Packet Loss (`sloss`)", min_value=0)
-        swin = st.number_input("Source TCP Window Size (`swin`)", min_value=0)
-        trans_depth = st.number_input("Transaction Depth (`trans_depth`)", min_value=0)
-        ct_flw_http_mthd = st.text_input("HTTP Method (`ct_flw_http_mthd`)", value="-")
-        is_ftp_login = st.selectbox("Is FTP Login (`is_ftp_login`)", [0, 1])
-        attack_cat = st.text_input("(Optional) Ground Truth Label (`attack_cat`)", value="")
-
-    submitted = st.form_submit_button("Run IDS Prediction")
-
-if submitted:
-    user_input = pd.DataFrame([{
-        "dur": dur,
-        "proto": proto,
-        "service": service,
-        "state": state,
-        "sbytes": sbytes,
-        "dbytes": dbytes,
-        "spkts": spkts,
-        "dpkts": dpkts,
-        "sttl": sttl,
-        "sload": sload,
-        "sloss": sloss,
-        "swin": swin,
-        "trans_depth": trans_depth,
-        "ct_flw_http_mthd": ct_flw_http_mthd,
-        "is_ftp_login": is_ftp_login,
-        "attack_cat": attack_cat
-    }])
-
-    features_to_drop, category_encodings, model = load_model_artifacts()
-    processed_input = preprocess_input(user_input, features_to_drop, category_encodings)
-
-    if processed_input is not None:
-        prediction = model.predict(processed_input)[0]
-        st.success(f"Prediction: {prediction}")
-        st.markdown("""
-        - **13** → Normal Traffic  
-        - **Other values** → Intrusion Category  
-        """)
-    else:
-        st.error("Preprocessing failed. Please check your input.")
+# ---------------------------
+# Load trained model and metadata
+# ---------------------------
+model = joblib.load('xgb_model.pkl')
+
+with open('features_to_drop.pkl', 'rb') as f:
+    features_to_drop = pickle.load(f)
+
+with open('category_encodings.pkl', 'rb') as f:
+    category_encodings = pickle.load(f)
+
+# ---------------------------
+# App UI
+# ---------------------------
+st.title("🚨 Intrusion Detection In Networks")
+st.markdown("Paste a single row of preprocessed NB15 features (comma-separated) below to predict the intrusion type.")
+
+user_input = st.text_area("Input (1 row from NB15_combined_preprocessed.csv):", height=150)
+
+if st.button("Predict"):
+    try:
+        # Convert user input string into a DataFrame
+        row = [x.strip() for x in user_input.strip().split(",")]
+        input_df = pd.DataFrame([row])
+
+        # Load a sample row to get correct column names and data types
+        sample_df = pd.read_csv('NB15_combined_preprocessed.csv', nrows=1)
+        columns = sample_df.drop('attack_cat', axis=1).columns.tolist()
+
+        # Ensure correct number of columns
+        if len(row) != len(columns):
+            st.error(f"Expected {len(columns)} values, but got {len(row)}.")
+        else:
+            input_df.columns = columns
+            # Convert to appropriate types
+            input_df = input_df.apply(pd.to_numeric, errors='coerce')
+
+            # Drop high-correlation features
+            input_df = input_df.drop(columns=features_to_drop, errors='ignore')
+
+            # Predict
+            prediction = model.predict(input_df)[0]
+            st.success(f"🛡️ Predicted Intrusion Category: **{prediction}**")
+    except Exception as e:
+        st.error(f"⚠️ Error processing input: {e}")