data_scaler_methods:
  StandardScaler:
    description: "标准化特征，使其均值为0，标准差为1。"
    parameters:
      - name: "copy"
        type: "bool"
        default: "True"
        description: "是否复制数据，若为False，则会对原始数据进行修改。"
      - name: "with_mean"
        type: "bool"
        default: "True"
        description: "是否去均值处理。如果为False，则不做均值化处理。"
      - name: "with_std"
        type: "bool"
        default: "True"
        description: "是否按标准差缩放。如果为False，则不做标准差处理。"

  MinMaxScaler:
    description: "将特征缩放到指定范围，通常是[0, 1]，保持原始数据比例。"
    parameters:
      - name: "feature_range"
        type: "tuple"
        default: "(0, 1)"
        description: "输出范围，控制转换后数据的最小值和最大值。"
      - name: "copy"
        type: "bool"
        default: "True"
        description: "是否复制数据，若为False，则会对原始数据进行修改。"
        
  RobustScaler:
    description: "使用中位数和四分位距进行缩放，适用于包含异常值的数据。"
    parameters:
      - name: "center"
        type: "bool"
        default: "True"
        description: "是否使用中位数进行中心化处理。如果为False，则不进行中心化。"
      - name: "scale"
        type: "bool"
        default: "True"
        description: "是否使用四分位距进行缩放。如果为False，则不进行缩放。"
      - name: "copy"
        type: "bool"
        default: "True"
        description: "是否复制数据，若为False，则会对原始数据进行修改。"

  Normalizer:
    description: "对样本（而非特征）进行归一化处理，使每个样本的范数为1。"
    parameters:
      - name: "norm"
        type: "str"
        default: "'l2'"
        description: "使用的归一化范数。'l1'、'l2'、'max'。'l2'表示L2范数（默认），'l1'表示L1范数，'max'表示最大值归一化。"
      - name: "copy"
        type: "bool"
        default: "True"
        description: "是否复制数据，若为False，则会对原始数据进行修改。"

  Binarizer:
    description: "将数据二值化，根据阈值将特征值大于该阈值的设为1，否则设为0。"
    parameters:
      - name: "threshold"
        type: "float"
        default: "0.0"
        description: "二值化的阈值。大于该值的样本会被设置为1，其他为0。"
      - name: "copy"
        type: "bool"
        default: "True"
        description: "是否复制数据，若为False，则会对原始数据进行修改。"


missing_value_handling_methods:
  SimpleImputer:
    description: "使用统计方法（如均值、中位数、众数）或常数值填充缺失值。"
    parameters:
      - name: "missing_values"
        type: "int, float, str, np.nan 或 None"
        default: "np.nan"
        description: "指定需要填充的缺失值。"
      - name: "strategy"
        type: "str"
        default: "'mean'"
        description: "填充策略，可选值为 'mean'（均值）、'median'（中位数）、'most_frequent'（众数）和 'constant'（常数）。"
      - name: "fill_value"
        type: "str 或 数值"
        default: "None"
        description: "当 strategy='constant' 时，指定用于填充的常数值。"
      - name: "verbose"
        type: "int"
        default: "0"
        description: "控制冗长度，0 表示不输出信息。"
      - name: "copy"
        type: "bool"
        default: "True"
        description: "是否创建数据的副本进行填充，False 则在原数据上进行填充。"
      - name: "add_indicator"
        type: "bool"
        default: "False"
        description: "是否添加缺失值指示器特征，标记缺失值的位置。"

  IterativeImputer:
    description: "使用多重插补方法，基于其他特征预测缺失值。"
    parameters:
      - name: "estimator"
        type: "对象"
        default: "BayesianRidge()"
        description: "用于预测的估计器对象。"
      - name: "missing_values"
        type: "数值"
        default: "np.nan"
        description: "表示缺失值的占位符。"
      - name: "max_iter"
        type: "int"
        default: "10"
        description: "插补过程的最大迭代次数。"
      - name: "tol"
        type: "float"
        default: "1e-3"
        description: "早停的容忍度。"
      - name: "n_nearest_features"
        type: "int"
        default: "None"
        description: "用于预测的最近特征数量。"
      - name: "initial_strategy"
        type: "str"
        default: "'mean'"
        description: "初始插补的策略。"
      - name: "imputation_order"
        type: "str"
        default: "'ascending'"
        description: "插补的顺序。"
      - name: "skip_complete"
        type: "bool"
        default: "False"
        description: "是否跳过没有缺失值的特征。"
      - name: "min_value"
        type: "float 或 array-like"
        default: "None"
        description: "每个特征的最小可接受值。"
      - name: "max_value"
        type: "float 或 array-like"
        default: "None"
        description: "每个特征的最大可接受值。"
      - name: "verbose"
        type: "int"
        default: "0"
        description: "控制冗长度。"
      - name: "random_state"
        type: "int, RandomState 实例或 None"
        default: "None"
        description: "随机数生成器的种子。"
      - name: "add_indicator"
        type: "bool"
        default: "False"
        description: "是否添加缺失值指示器特征。"

  KNNImputer:
    description: "基于k近邻算法，用相似样本的值填充缺失值。"
    parameters:
      - name: "missing_values"
        type: "数值"
        default: "np.nan"
        description: "表示缺失值的占位符。"
      - name: "n_neighbors"
        type: "int"
        default: "5"
        description: "用于插补的邻居数量。"
      - name: "weights"
        type: "str 或 callable"
        default: "'uniform'"
        description: "权重函数，可选 'uniform'、'distance' 或自定义函数。"
      - name: "metric"
        type: "str 或 callable"
        default: "'nan_euclidean'"
        description: "距离度量方式。"
      - name: "copy"
        type: "bool"
        default: "True"
        description: "是否创建数据的副本进行填充。"
      - name: "add_indicator"
        type: "bool"
        default: "False"
        description: "是否添加缺失值指示器特征。"

  MissingIndicator:
    description: "生成指示器变量，标记数据中缺失值的位置。"
    parameters:
      - name: "missing_values"
        type: "数值"
        default: "np.nan"
        description: "表示缺失值的占位符。"
      - name: "features"
        type: "str"
        default: "'missing-only'"
        description: "指示器特征的范围，可选 'missing-only' 或 'all'。"
      - name: "sparse"
        type: "bool"
        default: "False"
        description: "是否返回稀疏矩阵。"
      - name: "error_on_new"
        type: "bool"
        default: "True"
        description: "在 transform 时遇到新特征时是否报错。"


outlier_detection_methods:
  IsolationForest:
    description: "通过构建随机决策树，将数据分割以孤立异常点的算法。"
    parameters:
      - name: "n_estimators"
        type: "int"
        default: "100"
        description: "森林中树的数量。"
      - name: "max_samples"
        type: "int 或 float"
        default: "'auto'"
        description: "用于构建每棵树的样本数量。'auto' 表示使用数据集大小。"
      - name: "contamination"
        type: "float"
        default: "0.1"
        description: "数据集中异常点的比例，用于确定决策函数的阈值。"
      - name: "max_features"
        type: "int 或 float"
        default: "1.0"
        description: "用于构建每棵树的特征数量。"
      - name: "bootstrap"
        type: "bool"
        default: "False"
        description: "是否对样本进行有放回抽样。"
      - name: "n_jobs"
        type: "int"
        default: "None"
        description: "并行运行的作业数量。"
      - name: "random_state"
        type: "int, RandomState 实例或 None"
        default: "None"
        description: "控制随机性。"
      - name: "verbose"
        type: "int"
        default: "0"
        description: "控制冗余信息的输出。"

  OneClassSVM:
    description: "使用支持向量机寻找将正常数据与异常数据分离的超平面。"
    parameters:
      - name: "kernel"
        type: "str"
        default: "'rbf'"
        description: "核函数类型，如 'linear'、'poly'、'rbf' 等。"
      - name: "degree"
        type: "int"
        default: "3"
        description: "多项式核函数的度，仅在 kernel='poly' 时有效。"
      - name: "gamma"
        type: "float 或 'scale' 或 'auto'"
        default: "'scale'"
        description: "核系数。"
      - name: "coef0"
        type: "float"
        default: "0.0"
        description: "核函数中的独立项。"
      - name: "tol"
        type: "float"
        default: "1e-3"
        description: "停止标准的容忍度。"
      - name: "nu"
        type: "float"
        default: "0.5"
        description: "训练误差的上限和支持向量的下限。"
      - name: "shrinking"
        type: "bool"
        default: "True"
        description: "是否使用收缩启发式。"
      - name: "cache_size"
        type: "float"
        default: "200"
        description: "缓存大小（以 MB 为单位）。"
      - name: "verbose"
        type: "bool"
        default: "False"
        description: "启用详细输出。"
      - name: "max_iter"
        type: "int"
        default: "-1"
        description: "最大迭代次数，-1 表示无限制。"

  LocalOutlierFactor:
    description: "通过比较样本与其邻居的局部密度差异来识别异常点。"
    parameters:
      - name: "n_neighbors"
        type: "int"
        default: "20"
        description: "用于计算局部密度的邻居数量。"
      - name: "algorithm"
        type: "str"
        default: "'auto'"
        description: "用于最近邻搜索的算法。"
      - name: "leaf_size"
        type: "int"
        default: "30"
        description: "BallTree 或 KDTree 的叶子大小。"
      - name: "metric"
        type: "str 或 callable"
        default: "'minkowski'"
        description: "距离度量方式。"
      - name: "p"
        type: "int"
        default: "2"
        description: "Minkowski 度量的幂参数。"
      - name: "metric_params"
        type: "dict"
        default: "None"
        description: "度量函数的其他关键字参数。"
      - name: "contamination"
        type: "float 或 'auto'"
        default: "'auto'"
        description: "数据集中异常点的比例。"
      - name: "novelty"
        type: "bool"
        default: "False"
        description: "是否用于新颖性检测。"
      - name: "n_jobs"
        type: "int"
        default: "None"
        description: "并行运行的作业数量。"

  EllipticEnvelope:
    description: "假设数据服从高斯分布，拟合一个椭圆包络以包围数据，超出包络的点被视为异常。"
    parameters:
      - name: "store_precision"
        type: "bool"
        default: "True"
        description: "是否存储精度矩阵。"
      - name: "assume_centered"
        type: "bool"
        default: "False"
        description: "假设数据已中心化。"
      - name: "support_fraction"
        type: "float"
        default: "None"
        description: "用于估计协方差的样本比例。"
      - name: "contamination"
        type: "float"
        default: "0.1"
        description: "数据集中异常点的比例。"
      - name: "random_state"
        type: "int, RandomState 实例或 None"
        default: "None"
        description: "控制随机性。"