from fastapi import FastAPI
from pydantic import BaseModel
import numpy as np
import joblib


# load model
model = joblib.load("credit_risk_xgb.pkl")
scaler = joblib.load("credit_risk_scaler.pkl")

app = FastAPI()

class CreditRiskInput(BaseModel):
    age: int
    income: float
    loan_amount: float
    credit_score: int
    years_employed: int
    missed_payments: int

@app.get("/")
def home():
    return {"message": "API is running"}

@app.post("/predict")
def predict(data: CreditRiskInput):
    arr = np.array([[
        data.age,
        data.income,
        data.loan_amount,
        data.credit_score,
        data.years_employed,
        data.missed_payments
    ]])

    arr_scaled = scaler.transform(arr)

    prob = float(model.predict(arr_scaled)[0])
    pred = 1 if prob > 0.5 else 0

    return {
        "prediction": pred,
        "risk_label": "High Risk" if pred == 1 else "Low Risk",
        "default_probability": round(prob, 4)
    }

# -----------------------------
# Health check endpoint
# -----------------------------
@app.get("/")
def home():
    return {
        "message": "Credit Risk Prediction API is running"
    }

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

# -----------------------------
# Prediction endpoint
# -----------------------------
@app.post("/predict")
def predict_risk(data: CreditRiskInput):
    input_data = np.array([[
        data.age,
        data.income,
        data.loan_amount,
        data.credit_score,
        data.years_employed,
        data.missed_payments
    ]])

    input_scaled = scaler.transform(input_data)

    probability = float(model.predict_proba(input_scaled)[0][1])
    prediction = int(model.predict(input_scaled)[0])

    risk_label = "High Risk" if prediction == 1 else "Low Risk"

    return {
        "prediction": prediction,
        "risk_label": risk_label,
        "default_probability": round(probability, 4)
    }