|
|
"""
|
|
|
特征处理模块
|
|
|
包含特征数据结构定义、特征转换、权重应用和归一化等功能
|
|
|
"""
|
|
|
from typing import List, Dict, Any, Union, Optional
|
|
|
import os
|
|
|
import pandas as pd
|
|
|
import numpy as np
|
|
|
from sklearn.preprocessing import StandardScaler
|
|
|
import pickle
|
|
|
from pydantic import BaseModel, Field
|
|
|
import logging
|
|
|
import shutil
|
|
|
|
|
|
class Features(BaseModel):
|
|
|
"""
|
|
|
单个样本的特征数据结构
|
|
|
|
|
|
Attributes:
|
|
|
somatization: 躯体化症状得分
|
|
|
obsessive_compulsive: 强迫症状得分
|
|
|
interpersonal_sensitivity: 人际关系敏感得分
|
|
|
depression: 抑郁症状得分
|
|
|
anxiety: 焦虑症状得分
|
|
|
hostility: 敌对症状得分
|
|
|
terror: 恐怖症状得分
|
|
|
paranoia: 偏执症状得分
|
|
|
psychoticism: 精神病性得分
|
|
|
other: 其他症状得分
|
|
|
father_parenting_style: 父亲教养方式 (1: 温暖与理解, 0: 其他)
|
|
|
mother_parenting_style: 母亲教养方式 (1: 温暖与理解, 0: 其他)
|
|
|
self_assessed_family_economic_condition: 家庭经济状况 (2: 贫困, 1: 较差, 0: 其他)
|
|
|
history_of_psychological_counseling: 心理咨询史 (True: 有, False: 无)
|
|
|
absenteeism_above_average: 出勤情况 (True: 高于平均, False: 低于平均)
|
|
|
academic_warning: 学业预警 (True: 有预警, False: 无预警)
|
|
|
label: 分类标签 (0-3)
|
|
|
"""
|
|
|
# SCL评测量（范围0-4）
|
|
|
somatization: float = Field(..., description="躯体化症状得分")
|
|
|
obsessive_compulsive: float = Field(..., description="强迫症状得分")
|
|
|
interpersonal_sensitivity: float = Field(..., description="人际关系敏感得分")
|
|
|
depression: float = Field(..., description="抑郁症状得分")
|
|
|
anxiety: float = Field(..., description="焦虑症状得分")
|
|
|
hostility: float = Field(..., description="敌对症状得分")
|
|
|
terror: float = Field(..., description="恐怖症状得分")
|
|
|
paranoia: float = Field(..., description="偏执症状得分")
|
|
|
psychoticism: float = Field(..., description="精神病性得分")
|
|
|
other: float = Field(..., description="其他症状得分")
|
|
|
|
|
|
# 基本信息特征
|
|
|
father_parenting_style: int = Field(..., description="父亲教养方式")
|
|
|
mother_parenting_style: int = Field(..., description="母亲教养方式")
|
|
|
self_assessed_family_economic_condition: int = Field(..., description="家庭经济状况")
|
|
|
history_of_psychological_counseling: bool = Field(..., description="心理咨询史")
|
|
|
|
|
|
# 日常行为特征
|
|
|
absenteeism_above_average: bool = Field(..., description="出勤情况")
|
|
|
academic_warning: bool = Field(..., description="学业预警")
|
|
|
|
|
|
# 标签
|
|
|
label: int = Field(..., description="分类标签")
|
|
|
|
|
|
class FeatureProcessor:
|
|
|
"""
|
|
|
特征处理类，负责特征的转换和处理
|
|
|
|
|
|
Attributes:
|
|
|
config: 配置信息字典
|
|
|
feature_mapping: 特征映射规则
|
|
|
feature_names: 特征名称列表
|
|
|
"""
|
|
|
def __init__(self, config: Dict[str, Any]) -> None:
|
|
|
self.config = config
|
|
|
self.feature_mapping = config['features']['feature_mapping']
|
|
|
self.feature_names = config['features']['feature_names']
|
|
|
|
|
|
def _process_scl_features(self, features: Features) -> Dict[str, float]:
|
|
|
"""
|
|
|
处理SCL相关特征
|
|
|
|
|
|
Args:
|
|
|
features: 特征对象
|
|
|
|
|
|
Returns:
|
|
|
处理后的SCL特征字典
|
|
|
"""
|
|
|
return {
|
|
|
"somatization": features.somatization,
|
|
|
"obsessive_compulsive": features.obsessive_compulsive,
|
|
|
"interpersonal_sensitivity": features.interpersonal_sensitivity,
|
|
|
"depression": features.depression,
|
|
|
"anxiety": features.anxiety,
|
|
|
"hostility": features.hostility,
|
|
|
"terror": features.terror,
|
|
|
"paranoia": features.paranoia,
|
|
|
"psychoticism": features.psychoticism,
|
|
|
"other": features.other
|
|
|
}
|
|
|
|
|
|
def _calculate_multi_factor(self, scl_features: Dict[str, float]) -> float:
|
|
|
"""
|
|
|
计算多因子症状得分
|
|
|
|
|
|
Args:
|
|
|
scl_features: SCL特征字典
|
|
|
|
|
|
Returns:
|
|
|
多因子症状得分
|
|
|
"""
|
|
|
severe_symptoms_count = sum(1 for value in scl_features.values() if value > 3.0)
|
|
|
return severe_symptoms_count / len(scl_features)
|
|
|
|
|
|
def create_feature_df(self, features_list: List[Features]) -> pd.DataFrame:
|
|
|
"""
|
|
|
创建特征DataFrame
|
|
|
|
|
|
Args:
|
|
|
features_list: 特征对象列表
|
|
|
|
|
|
Returns:
|
|
|
处理后的特征DataFrame
|
|
|
"""
|
|
|
processed_features = []
|
|
|
|
|
|
for features in features_list:
|
|
|
scl_features = self._process_scl_features(features)
|
|
|
|
|
|
try:
|
|
|
feature_dict = {
|
|
|
'父亲教养方式数字化': self.feature_mapping['父亲教养方式'].get(
|
|
|
features.father_parenting_style,
|
|
|
self.feature_mapping['父亲教养方式'][0] # 默认值
|
|
|
),
|
|
|
'母亲教养方式数字化': self.feature_mapping['母亲教养方式'].get(
|
|
|
features.mother_parenting_style,
|
|
|
self.feature_mapping['母亲教养方式'][0] # 默认值
|
|
|
),
|
|
|
'自评家庭经济条件数字化': self.feature_mapping['家庭经济条件'].get(
|
|
|
features.self_assessed_family_economic_condition,
|
|
|
self.feature_mapping['家庭经济条件'][0] # 默认值
|
|
|
),
|
|
|
'有无心理治疗（咨询）史数字化': self.feature_mapping['心理咨询史'][features.history_of_psychological_counseling],
|
|
|
'强迫症状数字化': features.obsessive_compulsive / 4,
|
|
|
'人际关系敏感数字化': features.interpersonal_sensitivity / 4,
|
|
|
'抑郁数字化': features.depression / 4,
|
|
|
'多因子症状': self._calculate_multi_factor(scl_features),
|
|
|
'出勤情况数字化': self.feature_mapping['出勤情况'][features.absenteeism_above_average],
|
|
|
'学业情况数字化': self.feature_mapping['学业情况'][features.academic_warning],
|
|
|
self.config['features']['label_name']: features.label
|
|
|
}
|
|
|
processed_features.append(feature_dict)
|
|
|
except KeyError as e:
|
|
|
logging.error(f"Invalid feature value: {e}")
|
|
|
# 使用默认值继续处理
|
|
|
feature_dict = {
|
|
|
'父亲教养方式数字化': self.feature_mapping['父亲教养方式'][0],
|
|
|
'母亲教养方式数字化': self.feature_mapping['母亲教养方式'][0],
|
|
|
'自评家庭经济条件数字化': self.feature_mapping['家庭经济条件'][0],
|
|
|
'有无心理治疗（咨询）史数字化': self.feature_mapping['心理咨询史'][False],
|
|
|
'强迫症状数字化': features.obsessive_compulsive / 4,
|
|
|
'人际关系敏感数字化': features.interpersonal_sensitivity / 4,
|
|
|
'抑郁数字化': features.depression / 4,
|
|
|
'多因子症状': self._calculate_multi_factor(scl_features),
|
|
|
'出勤情况数字化': self.feature_mapping['出勤情况'][False],
|
|
|
'学业情况数字化': self.feature_mapping['学业情况'][False],
|
|
|
self.config['features']['label_name']: features.label
|
|
|
}
|
|
|
processed_features.append(feature_dict)
|
|
|
|
|
|
return pd.DataFrame(processed_features)
|
|
|
|
|
|
class FeatureWeightApplier:
|
|
|
"""特征权重应用类"""
|
|
|
@staticmethod
|
|
|
def apply_weights(df: pd.DataFrame, feature_names: List[str],
|
|
|
feature_weights: List[float]) -> pd.DataFrame:
|
|
|
"""应用特征权重"""
|
|
|
max_weight = max(feature_weights)
|
|
|
weights_scaled = [w / max_weight for w in feature_weights]
|
|
|
|
|
|
features_data = df[feature_names].values
|
|
|
labels = df['类别'].values
|
|
|
|
|
|
features_weighted = features_data * np.array(weights_scaled)
|
|
|
result_df = pd.DataFrame(features_weighted, columns=feature_names)
|
|
|
result_df['类别'] = labels
|
|
|
|
|
|
return result_df
|
|
|
|
|
|
class FeatureNormalizer:
|
|
|
"""特征归一化处理类"""
|
|
|
def __init__(self, config: Dict[str, Any]):
|
|
|
self.config = config
|
|
|
self.feature_groups = {}
|
|
|
self.scalers = {}
|
|
|
|
|
|
for group_name, features in config['features']['groups'].items():
|
|
|
indices = [config['features']['feature_names'].index(feature)
|
|
|
for feature in features]
|
|
|
self.feature_groups[group_name] = indices
|
|
|
|
|
|
def fit_transform(self, features_data: np.ndarray) -> np.ndarray:
|
|
|
"""训练时的归一化转换"""
|
|
|
features_scaled = features_data.copy()
|
|
|
for group_name, indices in self.feature_groups.items():
|
|
|
self.scalers[group_name] = StandardScaler()
|
|
|
features_scaled[:, indices] = self.scalers[group_name].fit_transform(features_data[:, indices])
|
|
|
return features_scaled
|
|
|
|
|
|
def transform(self, features_data: np.ndarray) -> np.ndarray:
|
|
|
"""预测时的归一化转换"""
|
|
|
features_scaled = features_data.copy()
|
|
|
for group_name, indices in self.feature_groups.items():
|
|
|
features_scaled[:, indices] = self.scalers[group_name].transform(features_data[:, indices])
|
|
|
return features_scaled
|
|
|
|
|
|
def save(self, path: str) -> None:
|
|
|
"""
|
|
|
保存归一化参数
|
|
|
|
|
|
Args:
|
|
|
path: 保存路径
|
|
|
"""
|
|
|
# 确保目录存在
|
|
|
os.makedirs(os.path.dirname(path), exist_ok=True)
|
|
|
|
|
|
with open(path, 'wb') as f:
|
|
|
pickle.dump(self.scalers, f)
|
|
|
|
|
|
def load(self, path: str) -> None:
|
|
|
"""
|
|
|
加载归一化参数
|
|
|
|
|
|
Args:
|
|
|
path: 加载路径
|
|
|
"""
|
|
|
# 如果主路径不存在，尝试从默认路径复制
|
|
|
if not os.path.exists(path):
|
|
|
default_path = os.path.join(
|
|
|
os.path.dirname(__file__),
|
|
|
'..',
|
|
|
self.config['paths']['normalizer']['default']
|
|
|
)
|
|
|
if os.path.exists(default_path):
|
|
|
# 确保目标目录存在
|
|
|
os.makedirs(os.path.dirname(path), exist_ok=True)
|
|
|
shutil.copyfile(default_path, path)
|
|
|
logging.info(f"Copied normalizer from default path: {default_path} to {path}")
|
|
|
else:
|
|
|
raise FileNotFoundError(f"No normalizer found in either path: {path} or {default_path}")
|
|
|
|
|
|
with open(path, 'rb') as f:
|
|
|
self.scalers = pickle.load(f)
|
|
|
|
|
|
def normalize_features(df: pd.DataFrame, feature_names: List[str],
|
|
|
is_train: bool, config: Dict[str, Any]) -> pd.DataFrame:
|
|
|
"""特征归一化处理入口函数"""
|
|
|
features_data = df[feature_names].values
|
|
|
labels = df[config['features']['label_name']].values
|
|
|
|
|
|
normalizer = FeatureNormalizer(config)
|
|
|
|
|
|
if is_train:
|
|
|
features_normalized = normalizer.fit_transform(features_data)
|
|
|
if config.get('paths', {}).get('normalizer', {}).get('train'):
|
|
|
normalizer.save(config['paths']['normalizer']['train'])
|
|
|
else:
|
|
|
normalizer.load(config['paths']['normalizer']['train'])
|
|
|
features_normalized = normalizer.transform(features_data)
|
|
|
|
|
|
result_df = pd.DataFrame(features_normalized, columns=feature_names)
|
|
|
result_df[config['features']['label_name']] = labels
|
|
|
|
|
|
return result_df |