app.py

import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt
import gradio as gr
import io
from PIL import Image

def eda_agent(file):
    try:
        df = pd.read_csv(file.name)
    except Exception as e:
        return f"❌ Error loading file: {e}", []

    output = ""

    # Step 1: Shape
    output += f"### 🧮 Dataset Shape: {df.shape}\n\n"

    # Step 2: Column Types
    output += "### 🔤 Column Data Types:\n" + df.dtypes.to_string() + "\n\n"

    # Step 3: Missing Values
    output += "### ❓ Missing Values:\n" + df.isnull().sum().to_string() + "\n\n"

    # Step 4: Duplicate Rows
    output += f"### 🔁 Duplicate Rows: {df.duplicated().sum()}\n\n"

    # Step 5: Summary Statistics
    output += "### 📊 Summary Statistics:\n" + df.describe(include='all').to_string() + "\n\n"

    # Step 6: First and Last 5 Rows
    output += "### 🔍 First 5 Rows:\n" + df.head().to_string() + "\n\n"
    output += "### 🔍 Last 5 Rows:\n" + df.tail().to_string() + "\n\n"

    # Step 7: Unique Values per Column
    output += "### 🧬 Unique Values per Column:\n"
    for col in df.columns:
        output += f"- {col}: {df[col].nunique()} unique values\n"
    output += "\n"

    # Step 8: Value Counts for Categorical Columns
    output += "### 🏷️ Top Value Counts (for categorical features):\n"
    cat_cols = df.select_dtypes(include='object').columns
    for col in cat_cols:
        output += f"\n#### {col}:\n{df[col].value_counts(dropna=False).head(5).to_string()}\n"

    # Plots
    plots = []

    # Step 9: Numeric Histograms
    numeric_cols = df.select_dtypes(include=['int64', 'float64']).columns
    for col in numeric_cols[:5]:  # Limit to first 5
        plt.figure(figsize=(6, 4))
        sns.histplot(df[col].dropna(), kde=True)
        plt.title(f'Distribution of {col}')
        buf = io.BytesIO()
        plt.savefig(buf, format='png')
        buf.seek(0)
        plots.append(Image.open(buf))
        plt.close()

    # Step 10: Correlation Heatmap
    if len(numeric_cols) >= 2:
        plt.figure(figsize=(10, 6))
        sns.heatmap(df[numeric_cols].corr(), annot=True, cmap='coolwarm')
        plt.title("Correlation Heatmap")
        buf = io.BytesIO()
        plt.savefig(buf, format='png')
        buf.seek(0)
        plots.append(Image.open(buf))
        plt.close()

    return output, plots

# Gradio interface
demo = gr.Interface(
    fn=eda_agent,
    inputs=gr.File(label="📁 Upload CSV File"),
    outputs=[
        gr.Textbox(label="📝 Full EDA Summary", lines=40),
        gr.Gallery(label="📊 Visualizations")
    ],
    title="📈 Auto EDA Agent",
    description="Upload your CSV to get full exploratory data analysis — summary, stats, and visualizations."
)

demo.launch()