Code Editor for app.py

import streamlit as st
import pandas as pd
import numpy as np
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LinearRegression
import plotly.express as px
from datetime import datetime, timedelta

# Set page title
st.title("Electricity Consumption Prediction")

# Generate sample data
def generate_sample_data():
    np.random.seed(42)
    dates = pd.date_range(start='2020-01-01', end='2023-12-31', freq='M')
    consumption = np.random.normal(1000, 200, len(dates))  # Random consumption values
    # Add seasonal variation
    seasonal_factor = np.sin(np.arange(len(dates)) * 2 * np.pi / 12) * 200
    consumption += seasonal_factor
    consumption = np.abs(consumption)  # Ensure positive values
    
    df = pd.DataFrame({
        'Date': dates,
        'Consumption': consumption,
        'Month': dates.month,
        'Year': dates.year
    })
    return df

# Load data
df = generate_sample_data()

# Prepare features for model
X = df.index.values.reshape(-1, 1)
y = df['Consumption'].values

# Train model
model = LinearRegression()
model.fit(X, y)

# Create prediction interface
st.subheader("Predict Future Consumption")

# Allow user to select future months
months_to_predict = st.slider("Select number of months to predict:", 1, 24, 6)

# Generate future dates
last_date = df['Date'].iloc[-1]
future_dates = pd.date_range(start=last_date + timedelta(days=31), 
                           periods=months_to_predict, 
                           freq='M')

# Make predictions
future_indices = np.arange(len(df), len(df) + months_to_predict).reshape(-1, 1)
predictions = model.predict(future_indices)

# Create future DataFrame
future_df = pd.DataFrame({
    'Date': future_dates,
    'Consumption': predictions,
    'Month': future_dates.month,
    'Year': future_dates.year
})

# Display predictions
st.subheader("Predicted Consumption")
st.write(future_df)

# Visualize historical data and predictions
fig = px.line(df, x='Date', y='Consumption', title='Historical and Predicted Electricity Consumption')
fig.add_scatter(x=future_df['Date'], y=future_df['Consumption'], 
               name='Predictions', line=dict(dash='dash'))

st.plotly_chart(fig)

# Show model performance metrics
st.subheader("Model Information")
col1, col2 = st.columns(2)

with col1:
    st.metric("Model R² Score", f"{model.score(X, y):.3f}")

with col2:
    avg_consumption = future_df['Consumption'].mean()
    st.metric("Average Predicted Consumption", f"{avg_consumption:.2f} kWh")

# Add explanatory notes
st.markdown("""
### Notes:
- This is a simplified prediction model using linear regression
- The model takes into account historical trends and seasonal patterns
- Predictions are based on past consumption patterns
- Actual consumption may vary due to various factors not considered in this simple model
""")