Code Editor for app.py

import streamlit as st
import pandas as pd
import numpy as np
from sklearn.linear_model import LinearRegression
from datetime import datetime, timedelta

st.title("Power Consumption Predictor")

# Input for CSV URL
url = st.text_input(
    "Enter CSV URL",
    value="/kaggle/input/electric-power-consumption-data-set",
    help="Enter the URL of your power consumption CSV file"
)

@st.cache_data
def load_data(url):
    try:
        # For demo purposes, generate sample data if URL is the default one
        if url == "/kaggle/input/electric-power-consumption-data-set":
            # Generate sample data
            dates = pd.date_range(start='2020-01-01', end='2023-12-31', freq='D')
            np.random.seed(42)
            consumption = np.random.normal(loc=30, scale=5, size=len(dates))
            consumption = consumption + np.sin(np.arange(len(dates)) * 2 * np.pi / 365) * 10  # Add seasonal pattern
            df = pd.DataFrame({
                'Date': dates,
                'KWH': consumption
            })
        else:
            df = pd.read_csv(url)
            
        return df
    except Exception as e:
        st.error(f"Error loading data: {str(e)}")
        return None

# Load data
data = load_data(url)

if data is not None:
    st.subheader("Data Preview")
    st.dataframe(data.head())
    
    # Input for current consumption
    current_kwh = st.number_input("Enter current kilowatt/hour consumption", min_value=0.0, value=25.0)
    
    # Input for prediction month
    prediction_month = st.selectbox(
        "Select month for prediction",
        ["January", "February", "March", "April", "May", "June",
         "July", "August", "September", "October", "November", "December"]
    )
    
    def train_model(data):
        # For demo data
        if 'Date' in data.columns:
            data['Month'] = pd.to_datetime(data['Date']).dt.month
            data['Year'] = pd.to_datetime(data['Date']).dt.year
            X = data[['Month', 'Year']].values
            y = data['KWH'].values
        else:
            st.warning("Please ensure your data has 'Date' and consumption columns")
            return None
            
        model = LinearRegression()
        model.fit(X, y)
        return model
    
    if st.button("Predict"):
        model = train_model(data)
        
        if model is not None:
            # Get current date info
            current_date = datetime.now()
            next_year = current_date.year + 1
            prediction_month_num = datetime.strptime(prediction_month, '%B').month
            
            # Make prediction
            prediction = model.predict([[prediction_month_num, next_year]])[0]
            
            # Display results
            st.subheader("Prediction Results")
            col1, col2 = st.columns(2)
            
            with col1:
                st.metric(
                    label="Current Consumption",
                    value=f"{current_kwh:.2f} KWH"
                )
            
            with col2:
                st.metric(
                    label=f"Predicted Consumption ({prediction_month} {next_year})",
                    value=f"{prediction:.2f} KWH",
                    delta=f"{prediction - current_kwh:.2f} KWH"
                )
            
            # Plot historical data and prediction if using demo data
            if 'Date' in data.columns:
                st.subheader("Historical Consumption Pattern")
                st.line_chart(data.set_index('Date')['KWH'])