app.py

import streamlit as st
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from sklearn.ensemble import IsolationForest
from sklearn.preprocessing import StandardScaler

# App title
st.title("📈 Anomaly Detection Tool")

# 🎯 Streamlit Tabs
tab1, tab2, tab3 = st.tabs(["📖 About", "📊 Dataset Overview", "🚨 Anomaly Detection"])

# About Tab
with tab1:
    st.write("""
    This app detects anomalies in time-series data using the Isolation Forest algorithm.
    Users can visualize detected anomalies.
    
    ### How It Works:
    - **Step 1**: Load a dataset (CSV format from the Numenta Anomaly Benchmark `realKnownCause` dataset)
    - **Step 2**: Standardize numerical values for better anomaly detection
    - **Step 3**: Apply **Isolation Forest** to identify outliers
    - **Step 4**: Visualize the detected anomalies in a time-series plot
    """)

# Load dataset
file_path = "ambient_temperature_system_failure.csv"
df = pd.read_csv(file_path)

# Dataset Overview Tab
with tab2:
    st.write("### Dataset Overview")
    st.write(df.head())
    
    # Convert timestamp column to datetime
    df['timestamp'] = pd.to_datetime(df['timestamp'], errors='coerce')
    df = df.dropna(subset=['timestamp'])
    df.set_index('timestamp', inplace=True)
    
    st.write("### Processed Dataset")
    st.write(df.head())

# Anomaly Detection Tab
with tab3:
    st.write("### Detect Anomalies in the Data")
    
    # Standardize the data
    scaler = StandardScaler()
    df['scaled_value'] = scaler.fit_transform(df[['value']])
    
    # Apply Isolation Forest
    contamination_level = st.slider("Select Contamination Level", 0.01, 0.1, 0.05, 0.01)
    model = IsolationForest(contamination=contamination_level, random_state=42)
    df['anomaly'] = model.fit_predict(df[['scaled_value']])
    df['anomaly'] = df['anomaly'].map({1: 0, -1: 1})  # Convert to binary (1: anomaly, 0: normal)
    
    # Allow user to set anomaly score threshold
    threshold = st.slider("Set Anomaly Score Threshold", -1.0, 1.0, 0.0, 0.01)
    df["anomaly_score"] = model.decision_function(df[["scaled_value"]])
    df["anomaly"] = df["anomaly_score"] < threshold
    
    # Plot results
    fig, ax = plt.subplots(figsize=(12, 6))
    ax.plot(df.index, df['value'], label='Value', color='blue')
    ax.scatter(df.index[df['anomaly'] == 1], df['value'][df['anomaly'] == 1], color='red', label='Anomaly', marker='o')
    ax.set_xlabel('Timestamp')
    ax.set_ylabel('Value')
    ax.set_title('Anomaly Detection in Time-Series Data')
    ax.legend()
    st.pyplot(fig)