app.py

import gradio as gr
import pandas as pd
from sklearn.datasets import make_classification
from sklearn.ensemble import RandomForestClassifier
from evidently.report import Report
from evidently.metrics import DataDriftTable


def run_monitoring():
    # Generate synthetic healthcare claims data
    X, y = make_classification(
        n_samples=2000,
        n_features=5,
        weights=[0.95, 0.05],  # fraud = rare
        random_state=42
    )

    df = pd.DataFrame(X, columns=[f"feature_{i}" for i in range(5)])
    df["fraud_flag"] = y

    # Train model on first 1500 samples
    model = RandomForestClassifier()
    model.fit(df.iloc[:1500, :-1], df.iloc[:1500, -1])

    # Create drift in production batch
    df_prod = df.iloc[1500:].copy()
    df_prod["feature_0"] = df_prod["feature_0"] * 3  # drift injected

    # Generate Evidently drift report
    report = Report(metrics=[DataDriftTable()])
    report.run(reference_data=df.iloc[:1500], current_data=df_prod)

    # Save to HTML
    report.save_html("monitoring_report.html")

    return "monitoring_report.html"


with gr.Blocks() as demo:
    gr.Markdown("# Healthcare Fraud Detection Model Monitoring\nThis example shows data drift monitoring using Evidently.")
    btn = gr.Button("Run Monitoring")
    file_output = gr.File()

    btn.click(fn=run_monitoring, outputs=file_output)

demo.launch(share=True)