main.py

# main.py  — FastAPI model service
# Run: uvicorn main:app --reload
# Requires: pip install fastapi uvicorn pydantic joblib pandas

from fastapi import FastAPI
from fastapi.middleware.cors import CORSMiddleware
from pydantic import BaseModel
from typing import List, Dict, Any
import pandas as pd
import joblib
import os

# Path to your saved pipeline from the notebook
MODEL_PATH = os.getenv("MODEL_PATH", "diabetes_prediction_model_20251007.pkl")
pipe = joblib.load(MODEL_PATH)

# Feature schema expected by the pipeline (must match training)
NUMERIC_FEATURES = ['age','alcohol_consumption_per_week', 'physical_activity_minutes_per_week',
                     'diet_score', 'bmi', 'cholesterol_total', 'insulin_level', 'map', 'glucose_fasting']

CATEGORICAL_FEATURES = ['gender', 'ethnicity', 'education_level', 'income_level', 'employment_status',
                         'smoking_status', 'family_history_diabetes', 'hypertension_history', 'cardiovascular_history']

ALL_FEATURES = NUMERIC_FEATURES + CATEGORICAL_FEATURES

class Batch(BaseModel):
    data: List[Dict[str, Any]]

app = FastAPI()
app.add_middleware(
    CORSMiddleware,
    allow_origins=["*"], allow_credentials=True,
    allow_methods=["*"], allow_headers=["*"],
)

@app.get("/health")
def health():
    return {"status": "ok"}

@app.get("/features")
def features():
    return {"features": ALL_FEATURES}

@app.post("/predict_batch")
def predict_batch(batch: Batch):
    X = pd.DataFrame(batch.data)
    # Ensure all expected columns exist
    for col in ALL_FEATURES:
        if col not in X.columns:
            X[col] = 0
    # Reorder and type-hints (numerics safe-cast, categoricals to str)
    X = X[ALL_FEATURES].copy()
    for col in NUMERIC_FEATURES:
        X[col] = pd.to_numeric(X[col], errors="coerce").fillna(0)
    for col in CATEGORICAL_FEATURES:
        X[col] = X[col].astype(str).fillna("Unknown")
    probs = pipe.predict_proba(X)[:, 1].tolist()
    return {"probabilities": probs}