import joblib
from sklearn.pipeline import Pipeline
from sklearn.preprocessing import StandardScaler
from sklearn.ensemble import RandomForestClassifier

# 파이프라인 포함 모델 학습
pipe = Pipeline([
    ("scaler", StandardScaler()),
    ("model", RandomForestClassifier(n_estimators=100, random_state=42))
])
pipe.fit(X_train, y_train)

# 모델 저장 (파이프라인 전체 저장)
joblib.dump(pipe, "model/classifier_pipeline_v1.joblib")

# 모델 로드 및 예측
loaded_pipe = joblib.load("model/classifier_pipeline_v1.joblib")
predictions = loaded_pipe.predict(X_test)
print(f"로드된 모델 정확도: {loaded_pipe.score(X_test, y_test):.4f}")

from skl2onnx import convert_sklearn
from skl2onnx.common.data_types import FloatTensorType

# scikit-learn 모델을 ONNX로 변환
initial_type = [("float_input", FloatTensorType([None, X_train.shape[1]]))]
onnx_model = convert_sklearn(pipe, initial_types=initial_type)

# ONNX 파일 저장
with open("model/classifier_v1.onnx", "wb") as f:
    f.write(onnx_model.SerializeToString())

# ONNX Runtime으로 추론
import onnxruntime as rt
import numpy as np

sess = rt.InferenceSession("model/classifier_v1.onnx")
input_name = sess.get_inputs()[0].name
pred = sess.run(None, {input_name: X_test.astype(np.float32)})[0]
print(f"ONNX 추론 결과: {pred[:5]}")

import json
from datetime import datetime

# 모델 메타데이터 저장
metadata = {
    "version": "1.0.0",
    "created_at": datetime.now().isoformat(),
    "model_type": "RandomForestClassifier",
    "n_features": X_train.shape[1],
    "training_score": float(pipe.score(X_train, y_train)),
    "test_score": float(pipe.score(X_test, y_test)),
    "hyperparameters": {
        "n_estimators": 100,
        "max_depth": None,
    }
}

with open("model/metadata_v1.json", "w") as f:
    json.dump(metadata, f, indent=2, ensure_ascii=False)