app.py

# app.py
import gradio as gr
import pandas as pd
import joblib

# Load trained model
model = joblib.load("exoplanet_model.pkl")

# Features (must match training order)
FEATURES = ["orbital_period", "transit_duration", "planet_radius", "star_temp", "star_radius"]

# Mapping labels
LABELS = {1: "✅ Confirmed Exoplanet", 0: "🟡 Candidate", -1: "❌ False Positive"}

# ---- Single Prediction ----
def predict_single(orbital_period, transit_duration, planet_radius, star_temp, star_radius):
    df = pd.DataFrame([[orbital_period, transit_duration, planet_radius, star_temp, star_radius]], 
                      columns=FEATURES)
    pred = model.predict(df)[0]
    return LABELS.get(pred, "Unknown")

# ---- Bulk Prediction ----
def predict_bulk(file):
    df = pd.read_csv(file.name)
    
    # Ensure required columns exist
    missing_cols = [col for col in FEATURES if col not in df.columns]
    if missing_cols:
        return f"Missing columns in uploaded file: {missing_cols}", None
    
    preds = model.predict(df[FEATURES])
    df["prediction"] = [LABELS.get(p, "Unknown") for p in preds]
    
    # Save results
    output_file = "bulk_predictions.csv"
    df.to_csv(output_file, index=False)
    return df.head(10), output_file   # Show preview + downloadable file

# ---- Gradio UI ----
with gr.Blocks() as demo:
    gr.Markdown("## 🔭 Exoplanet Classifier – NASA Space Apps Challenge 2025")

    with gr.Tab("Single Prediction"):
        inputs = [
            gr.Number(label="Orbital Period (days)"),
            gr.Number(label="Transit Duration (hrs)"),
            gr.Number(label="Planet Radius (Earth radii)"),
            gr.Number(label="Star Temperature (K)"),
            gr.Number(label="Star Radius (Solar radii)")
        ]
        output = gr.Label(label="Prediction")
        gr.Interface(fn=predict_single, inputs=inputs, outputs=output).render()

    with gr.Tab("Bulk CSV Prediction"):
        file_input = gr.File(label="Upload CSV", file_types=[".csv"])
        preview = gr.Dataframe(label="Preview (first 10 rows)")
        download = gr.File(label="Download Predictions")
        btn = gr.Button("Run Bulk Classification")
        btn.click(fn=predict_bulk, inputs=file_input, outputs=[preview, download])

if __name__ == "__main__":
    demo.launch()