111/aicodes/module2_predictor/train_and_save.py

# -*- coding: utf-8 -*-
"""
CardioAI - Module 2: 模型训练与保存
一次性脚本 - 训练XGBoost模型并保存Pipeline
"""

import pandas as pd
import numpy as np
import joblib
from pathlib import Path
from sklearn.model_selection import train_test_split
from sklearn.compose import ColumnTransformer
from sklearn.preprocessing import StandardScaler, OneHotEncoder
from sklearn.pipeline import Pipeline
from sklearn.metrics import classification_report, confusion_matrix, roc_auc_score
import xgboost as xgb

# ============================================
# 配置与常量
# ============================================
CODE_ROOT = Path(r"E:\project_ai\claude_project1\aicodes")
DATA_PATH = Path(r"E:\project_ai\claude_project1\data\心血管疾病.xlsx")
MODEL_SAVE_PATH = CODE_ROOT / "module2_predictor" / "cardio_predictor_model.pkl"

# 类别映射字典
CHOLESTEROL_MAP = {1: "正常", 2: "高于正常", 3: "远高于正常"}
GLUC_MAP = {1: "正常", 2: "高于正常", 3: "远高于正常"}
GENDER_MAP = {1: "女性", 2: "男性"}


# ============================================
# 数据加载与清洗函数
# ============================================
def load_data():
    """加载心血管疾病数据"""
    print("正在加载数据...")
    df = pd.read_excel(DATA_PATH, engine='openpyxl')
    print(f"数据加载完成: {df.shape[0]} 条记录, {df.shape[1]} 个特征")
    return df


def clean_and_process_data(df):
    """
    数据清洗与特征工程

    处理步骤:
    1. 将age(天)转换为年
    2. 计算BMI
    3. 删除血压异常值
    4. 类别转换
    5. 删除id和原始age字段
    """
    print("\n开始数据清洗与特征工程...")

    df_clean = df.copy()

    # 1. 年龄转换: 天 -> 年 (四舍五入)
    df_clean['age_years'] = (df_clean['age'] / 365.25).round().astype(int)

    # 2. 计算BMI: weight / (height/100)^2
    df_clean['bmi'] = df_clean['weight'] / ((df_clean['height'] / 100) ** 2)

    # 3. 异常值处理
    original_count = len(df_clean)

    # 删除舒张压 >= 收缩压的记录
    df_clean = df_clean[df_clean['ap_lo'] < df_clean['ap_hi']]

    # 删除血压极端异常值 (收缩压范围: [90, 250], 舒张压范围: [60, 150])
    df_clean = df_clean[
        (df_clean['ap_hi'] >= 90) & (df_clean['ap_hi'] <= 250) &
        (df_clean['ap_lo'] >= 60) & (df_clean['ap_lo'] <= 150)
    ]

    removed_count = original_count - len(df_clean)
    print(f"  - 删除异常值: {removed_count} 条记录")

    # 4. 删除id和原始age字段
    df_clean = df_clean.drop(columns=['id', 'age'], errors='ignore')

    print(f"数据清洗完成: {len(df_clean)} 条有效记录")

    return df_clean


# ============================================
# 模型训练函数
# ============================================
def train_model(df):
    """
    训练XGBoost分类模型

    Args:
        df (pd.DataFrame): 清洗后的数据

    Returns:
        Pipeline: 包含预处理器和模型的完整Pipeline
    """
    print("\n开始模型训练...")

    # 定义特征列
    # 连续特征
    numeric_features = ['age_years', 'height', 'weight', 'bmi', 'ap_hi', 'ap_lo']

    # 分类特征
    categorical_features = ['gender', 'cholesterol', 'gluc', 'smoke', 'alco', 'active']

    # 特征与目标变量
    X = df[numeric_features + categorical_features]
    y = df['cardio']

    # 划分训练集和测试集
    X_train, X_test, y_train, y_test = train_test_split(
        X, y, test_size=0.2, random_state=42, stratify=y
    )

    print(f"  - 训练集: {X_train.shape[0]} 条记录")
    print(f"  - 测试集: {X_test.shape[0]} 条记录")

    # 构建预处理器
    preprocessor = ColumnTransformer(
        transformers=[
            ('num', StandardScaler(), numeric_features),
            ('cat', OneHotEncoder(drop='first', handle_unknown='ignore'), categorical_features)
        ],
        remainder='passthrough'
    )

    # 构建完整Pipeline
    pipeline = Pipeline(steps=[
        ('preprocessor', preprocessor),
        ('classifier', xgb.XGBClassifier(
            n_estimators=100,
            max_depth=6,
            learning_rate=0.1,
            subsample=0.8,
            colsample_bytree=0.8,
            random_state=42,
            eval_metric='logloss',
            use_label_encoder=False
        ))
    ])

    # 训练模型
    print("  - 正在训练 XGBoost 模型...")
    pipeline.fit(X_train, y_train)

    # 评估模型
    print("\n模型评估:")
    y_pred = pipeline.predict(X_test)
    y_pred_proba = pipeline.predict_proba(X_test)[:, 1]

    print("\n分类报告:")
    print(classification_report(y_test, y_pred, target_names=['无疾病', '有疾病']))

    print("\n混淆矩阵:")
    cm = confusion_matrix(y_test, y_pred)
    print(f"  真阴性(TN): {cm[0,0]}  假阳性(FP): {cm[0,1]}")
    print(f"  假阴性(FN): {cm[1,0]}  真阳性(TP): {cm[1,1]}")

    auc_score = roc_auc_score(y_test, y_pred_proba)
    print(f"\nAUC-ROC: {auc_score:.4f}")

    return pipeline


# ============================================
# 模型保存函数
# ============================================
def save_model(pipeline, save_path):
    """
    保存训练好的模型

    Args:
        pipeline (Pipeline): 训练好的Pipeline
        save_path (Path): 模型保存路径
    """
    print(f"\n正在保存模型到: {save_path}")
    joblib.dump(pipeline, save_path)
    print("模型保存完成!")


# ============================================
# 主程序
# ============================================
def main():
    """主程序入口"""
    print("=" * 60)
    print("CardioAI - Module 2: XGBoost模型训练与保存")
    print("=" * 60)

    # 1. 加载数据
    df = load_data()

    # 2. 数据清洗与特征工程
    df_clean = clean_and_process_data(df)

    # 3. 训练模型
    pipeline = train_model(df_clean)

    # 4. 保存模型
    save_model(pipeline, MODEL_SAVE_PATH)

    print("\n" + "=" * 60)
    print("训练完成! 模型已保存至:")
    print(f"  {MODEL_SAVE_PATH}")
    print("=" * 60)


if __name__ == "__main__":
    main()