app.py

import streamlit as st
import pandas as pd
import plotly.graph_objects as go
import pickle
from statsmodels.tsa.arima.model import ARIMA
import os
from datetime import datetime, timedelta

# Page configuration with custom theme
st.set_page_config(
    page_title="ERROR Log Analytics Dashboard",
    layout="wide",
    initial_sidebar_state="expanded"
)

# Custom CSS to enhance the UI
st.markdown("""
<style>
    .main-header {
        font-size: 2.5rem;
        font-weight: 700;
        color: #1E3A8A;
        margin-bottom: 1rem;
    }
    .sub-header {
        font-size: 1.5rem;
        font-weight: 600;
        color: #2563EB;
        margin-top: 2rem;
    }
    .card {
        background-color: #F8FAFC;
        border-radius: 0.5rem;
        padding: 1.5rem;
        box-shadow: 0 1px 3px rgba(0, 0, 0, 0.1);
        margin-bottom: 1rem;
    }
    .success-msg {
        background-color: #DCFCE7;
        color: #166534;
        padding: 0.75rem;
        border-radius: 0.375rem;
        border-left: 4px solid #16A34A;
        margin: 1rem 0;
    }
    .stButton>button {
        background-color: #2563EB;
        color: white;
        border: none;
        border-radius: 0.375rem;
        padding: 0.5rem 1rem;
    }
    .metrics-container {
        display: flex;
        justify-content: space-between;
    }
    .metric-card {
        background-color: #F0F9FF;
        border-radius: 0.5rem;
        padding: 1rem;
        text-align: center;
        box-shadow: 0 1px 2px rgba(0, 0, 0, 0.05);
        width: 32%;
    }
</style>
""", unsafe_allow_html=True)

# Sidebar for controls and information
with st.sidebar:
    st.image("https://www.svgrepo.com/show/374111/log.svg", width=100)
    st.markdown("## ERROR Log Analytics")
    st.markdown("---")
    st.markdown("### Model Configuration")

    # File selection (keeping default path but making it look configurable)
    file_path = st.text_input("Log file path", value="augmented_logs.csv")

    # ARIMA parameters
    st.markdown("### ARIMA Parameters")
    p = st.slider("Auto-regression (p)", min_value=0, max_value=5, value=1)
    d = st.slider("Differencing (d)", min_value=0, max_value=2, value=1)
    q = st.slider("Moving Average (q)", min_value=0, max_value=5, value=1)

    # Forecast range
    st.markdown("### Forecast Settings")
    forecast_days = st.slider("Forecast Horizon (days)", min_value=1, max_value=30, value=7)

    st.markdown("---")
    st.markdown("*This dashboard analyzes ERROR logs and forecasts future error rates using ARIMA modeling.*")

# Main content
st.markdown('<div class="main-header">📊 ERROR Log Analysis & Forecasting</div>', unsafe_allow_html=True)

if not os.path.exists(file_path):
    st.error(f"❌ File not found: {file_path}")
else:
    try:
        # Main app container
        with st.container():
            # Load and prep data
            df = pd.read_csv(file_path)
            df['timestamp'] = pd.to_datetime(df['timestamp'])
            df['date'] = df['timestamp'].dt.date

            # Group by date and count ERRORs
            daily_errors = df[df['log_level'] == 'ERROR'].groupby('date').size()
            daily_errors_ts = daily_errors.asfreq('D').fillna(0)

            # Display key metrics
            st.markdown('<div class="sub-header">Key Metrics</div>', unsafe_allow_html=True)

            col1, col2, col3 = st.columns(3)
            with col1:
                st.markdown('<div class="card">', unsafe_allow_html=True)
                st.metric("Total ERRORs", f"{int(daily_errors_ts.sum())}")
                st.markdown('</div>', unsafe_allow_html=True)

            with col2:
                st.markdown('<div class="card">', unsafe_allow_html=True)
                st.metric("Average Daily ERRORs", f"{daily_errors_ts.mean():.2f}")
                st.markdown('</div>', unsafe_allow_html=True)

            with col3:
                st.markdown('<div class="card">', unsafe_allow_html=True)
                if len(daily_errors_ts) >= 7:
                    last_week = daily_errors_ts[-7:].mean()
                    previous_week = daily_errors_ts[-14:-7].mean()
                    delta = ((last_week - previous_week) / previous_week * 100) if previous_week > 0 else 0
                    st.metric("7-Day Trend", f"{last_week:.2f}", f"{delta:.1f}%")
                else:
                    st.metric("7-Day Trend", "Insufficient data")
                st.markdown('</div>', unsafe_allow_html=True)

            # Historical data visualization
            st.markdown('<div class="sub-header">Historical ERROR Trends</div>', unsafe_allow_html=True)
            st.markdown('<div class="card">', unsafe_allow_html=True)

            chart_tab1, chart_tab2 = st.tabs(["Line Chart", "Bar Chart"])

            with chart_tab1:
                st.line_chart(daily_errors_ts)

            with chart_tab2:
                st.bar_chart(daily_errors_ts)

            st.markdown('</div>', unsafe_allow_html=True)

            # Model training
            st.markdown('<div class="sub-header">ARIMA Model Training</div>', unsafe_allow_html=True)
            st.markdown('<div class="card">', unsafe_allow_html=True)

            try:
                with st.spinner("Training ARIMA model..."):
                    # Use the parameters from sidebar
                    model = ARIMA(daily_errors_ts, order=(p, d, q))
                    model_fit = model.fit()

                    # Save model
                    with open("arima_model.pkl", "wb") as f:
                        pickle.dump(model_fit, f)

                    st.markdown('<div class="success-msg">✅ ARIMA model trained and saved successfully!</div>',
                                unsafe_allow_html=True)

                    # Display model summary in expander
                    with st.expander("View Model Details"):
                        st.code(str(model_fit.summary()))

            except Exception as arima_error:
                st.error(f"⚠️ ARIMA training failed: {arima_error}")

            st.markdown('</div>', unsafe_allow_html=True)

            # Forecast visualization
            st.markdown('<div class="sub-header">ERROR Forecast Analysis</div>', unsafe_allow_html=True)
            st.markdown('<div class="card">', unsafe_allow_html=True)

            forecast = model_fit.forecast(steps=forecast_days)
            forecast_dates = pd.date_range(start=daily_errors_ts.index[-1] + pd.Timedelta(days=1),
                                           periods=forecast_days)

            # Create forecast dataframe for additional analysis
            forecast_df = pd.DataFrame({
                'date': forecast_dates,
                'forecast': forecast.values,
                # Use a fixed standard deviation estimate instead of se_mean
                'lower_ci': forecast.values - 2 * forecast.values.std() if len(forecast) > 1 else forecast.values,
                'upper_ci': forecast.values + 2 * forecast.values.std() if len(forecast) > 1 else forecast.values * 1.2
            })

            # Round negative values to 0 for logical consistency
            forecast_df['lower_ci'] = forecast_df['lower_ci'].clip(lower=0)
            forecast_df['forecast'] = forecast_df['forecast'].clip(lower=0)

            # Enhanced plotly visualization
            fig = go.Figure()

            # Historical data
            fig.add_trace(go.Scatter(
                x=daily_errors_ts.index,
                y=daily_errors_ts.values,
                mode='lines+markers',
                name='Historical ERRORs',
                line=dict(color='#3B82F6', width=2)
            ))

            # Forecast
            fig.add_trace(go.Scatter(
                x=forecast_df['date'],
                y=forecast_df['forecast'],
                mode='lines+markers',
                name='Forecasted ERRORs',
                line=dict(color='#EF4444', width=2, dash='dash')
            ))

            # Confidence interval
            fig.add_trace(go.Scatter(
                x=forecast_df['date'].tolist() + forecast_df['date'].tolist()[::-1],
                y=forecast_df['upper_ci'].tolist() + forecast_df['lower_ci'].tolist()[::-1],
                fill='toself',
                fillcolor='rgba(239, 68, 68, 0.1)',
                line=dict(color='rgba(0,0,0,0)'),
                hoverinfo='skip',
                showlegend=False
            ))

            fig.update_layout(
                title='ERROR Log Forecast with Confidence Intervals',
                xaxis_title='Date',
                yaxis_title='ERROR Count',
                hovermode='x unified',
                legend=dict(x=0.01, y=0.99),
                template='plotly_white',
                height=500
            )

            st.plotly_chart(fig, use_container_width=True)

            # Forecast data table
            with st.expander("View Forecast Data"):
                forecast_df['date'] = forecast_df['date'].dt.date
                forecast_df['forecast'] = forecast_df['forecast'].round(2)
                forecast_df['lower_ci'] = forecast_df['lower_ci'].round(2)
                forecast_df['upper_ci'] = forecast_df['upper_ci'].round(2)
                st.dataframe(forecast_df)

            # Download forecast as CSV
            csv = forecast_df.to_csv(index=False)
            st.download_button(
                label="Download Forecast CSV",
                data=csv,
                file_name=f"error_forecast_{datetime.now().strftime('%Y%m%d')}.csv",
                mime="text/csv"
            )

            st.markdown('</div>', unsafe_allow_html=True)

    except Exception as e:
        st.error(f"❌ Error processing data: {e}")