app.py

import gradio as gr
import pandas as pd
import plotly.graph_objects as go
import numpy as np

def plot_zip_code_correlation(zip_codes_str, start_date, end_date):
    # Validate dates
    start_year = pd.to_datetime(start_date).year
    end_year = pd.to_datetime(end_date).year
    if start_year < 2000 or end_year < 2000:
        raise ValueError("Please select dates no earlier than the year 2000.")
    if start_year > end_year:
        raise ValueError("Start date must be before end date.")

    # Process ZIP codes (ensure 5-digit format)
    zip_codes = [z.strip().zfill(5) for z in zip_codes_str.split(",")]

    # Load data
    df = pd.read_csv('https://files.zillowstatic.com/research/public_csvs/zhvi/Zip_zhvi_uc_sfrcondo_tier_0.33_0.67_sm_sa_month.csv')

    # Ensure ZIP codes in dataframe are strings with leading zeros
    df['RegionName'] = df['RegionName'].astype(str).str.zfill(5)
    df = df[df['RegionName'].isin(zip_codes)]

    if df.empty:
        raise ValueError("No data found for the provided ZIP codes.")

    # Extract date columns within the selected range
    date_columns = []
    for col in df.columns[7:]:
        try:
            date = pd.to_datetime(col)
            if start_date <= str(date.date()) <= end_date:
                date_columns.append(col)
        except:
            continue

    if not date_columns:
        raise ValueError("No data available within the selected date range.")

    # Build price matrix
    price_matrix = []
    valid_zip_list = []

    for zip_code in zip_codes:
        df_zip = df[df['RegionName'] == zip_code]
        if not df_zip.empty:
            prices = df_zip.loc[:, date_columns].values.flatten()
            if not np.isnan(prices).all():
                price_matrix.append(prices)
                valid_zip_list.append(zip_code)

    if len(price_matrix) < 2:
        raise ValueError(f"Not enough data for correlation calculation. Ensure at least two valid ZIP codes with overlapping data between {start_date} and {end_date}.")

    price_matrix_df = pd.DataFrame(price_matrix, index=valid_zip_list, columns=date_columns)
    price_matrix_df = price_matrix_df.T.dropna()

    # Calculate correlation matrix
    corr_matrix = price_matrix_df.corr()

    # Prepare 3D plot
    z_data = corr_matrix.values
    x_data, y_data = np.meshgrid(valid_zip_list, valid_zip_list)

    fig = go.Figure(data=[go.Surface(z=z_data, x=x_data, y=y_data)])
    fig.update_layout(
        title=f'3D Correlation Matrix of Housing Prices ({start_date} to {end_date})',
        scene=dict(
            xaxis_title='ZIP Code',
            yaxis_title='ZIP Code',
            zaxis_title='Correlation',
        ),
        autosize=True
    )

    return fig

iface = gr.Interface(
    fn=plot_zip_code_correlation,
    inputs=[
        gr.Textbox(label="Enter comma-separated ZIP codes (e.g., 07001,07002,07003)"),
        gr.Textbox(label="Start Date (YYYY-MM-DD) - No earlier than 2000"),
        gr.Textbox(label="End Date (YYYY-MM-DD) - No earlier than 2000")
    ],
    outputs=gr.Plot(),
    title="3D ZIP Code Housing Price Correlation Matrix"
)

iface.launch(share=False, debug=True)