C:\编程软件\python\python.exe C:\Users\31864\Desktop\math\a.py 2025-05-22 20:02:10.584193: I tensorflow/core/util/port.cc:153] oneDNN custom operations are on. You may see slightly different numerical results due to floating-point round-off errors from different computation orders. To turn them off, set the environment variable `TF_ENABLE_ONEDNN_OPTS=0`. 2025-05-22 20:02:11.251475: I tensorflow/core/util/port.cc:153] oneDNN custom operations are on. You may see slightly different numerical results due to floating-point round-off errors from different computation orders. To turn them off, set the environment variable `TF_ENABLE_ONEDNN_OPTS=0`. Training model... Traceback (most recent call last): File "C:\Users\31864\Desktop\math\a.py", line 185, in <module> lstm_model.train_model() File "C:\Users\31864\Desktop\math\a.py", line 93, in train_model grid_result = grid_search.fit( ^^^^^^^^^^^^^^^^ File "C:\编程软件\python\Lib\site-packages\sklearn\base.py", line 1389, in wrapper return fit_method(estimator, *args, **kwargs) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\编程软件\python\Lib\site-packages\sklearn\model_selection\_search.py", line 933, in fit cv_orig = check_cv(self.cv, y, classifier=is_classifier(estimator)) ^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\编程软件\python\Lib\site-packages\sklearn\base.py", line 1237, in is_classifier return get_tags(estimator).estimator_type == "classifier" ^^^^^^^^^^^^^^^^^^^ File "C:\编程软件\python\Lib\site-packages\sklearn\utils\_tags.py", line 430, in get_tags sklearn_tags_provider[klass] = klass.__sklearn_tags__(estimator) # type: ignore[attr-defined] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "C:\编程软件\python\Lib\site-packages\sklearn\base.py", line 613, in __sklearn_tags__ tags = super().__sklearn_tags__() ^^^^^^^^^^^^^^^^^^^^^^^^ AttributeError: 'super' object has no attribute '__sklearn_tags__'

从你的错误信息来看，问题出在 `grid_search.fit()` 方法调用时，`sklearn` 的某些内部方法无法正确识别模型的标签（tags）。这通常是因为自定义模型没有正确继承 `BaseEstimator` 和/或 `ClassifierMixin` 或 `RegressorMixin`，导致 `sklearn` 无法正确处理模型。 以下是一个解决方案，确保你的模型正确继承了 `BaseEstimator` 和适当的 Mixin，并且实现了必要的方法。 ### 解决方案 假设你正在使用一个自定义的 LSTM 模型进行训练，以下是修正后的代码示例： ```python from sklearn.base import BaseEstimator, ClassifierMixin from tensorflow.keras.models import Sequential from tensorflow.keras.layers import LSTM, Dense from sklearn.model_selection import GridSearchCV import numpy as np class LstmModel(BaseEstimator, ClassifierMixin): def __init__(self, units=50, dropout=0.2, optimizer='adam'): self.units = units self.dropout = dropout self.optimizer = optimizer self.model = None def build_model(self, input_shape): model = Sequential() model.add(LSTM(units=self.units, return_sequences=True, input_shape=input_shape)) model.add(LSTM(units=self.units)) model.add(Dense(1, activation='sigmoid')) model.compile(optimizer=self.optimizer, loss='binary_crossentropy', metrics=['accuracy']) return model def fit(self, X, y, **kwargs): if self.model is None: self.model = self.build_model(input_shape=(X.shape[1], X.shape[2])) self.model.fit(X, y, **kwargs) return self def predict(self, X): return (self.model.predict(X) > 0.5).astype("int32") def score(self, X, y): _, accuracy = self.model.evaluate(X, y, verbose=0) return accuracy # 示例数据 X = np.random.rand(100, 10, 1) # 假设有100个样本，每个样本有10个时间步长和1个特征 y = np.random.randint(0, 2, size=(100,)) # 网格搜索参数 param_grid = { 'units': [50, 100], 'dropout': [0.2, 0.5], 'optimizer': ['adam', 'rmsprop'] } lstm_model = LstmModel() grid_search = GridSearchCV(estimator=lstm_model, param_grid=param_grid, cv=3, scoring='accuracy') grid_result = grid_search.fit(X, y) print("Best parameters:", grid_search.best_params_) print("Best score:", grid_search.best_score_) ``` ### 代码解释 1. **继承 `BaseEstimator` 和 `ClassifierMixin`**: - `BaseEstimator` 提供了 `get_params` 和 `set_params` 方法，这是 `GridSearchCV` 所需的基本功能。 - `ClassifierMixin` 提供了默认的 `score` 方法，用于评估分类器的性能。 2. **实现必要的方法**: - `fit`: 训练模型。 - `predict`: 预测输出。 - `score`: 计算模型的准确率。 3. **构建模型**: - 在 `build_model` 方法中定义了 LSTM 模型结构，并编译模型。 4. **网格搜索**: - 使用 `GridSearchCV` 进行超参数优化。 ### 错误原因分析 - 如果你的自定义模型没有继承 `BaseEstimator`，`sklearn` 的方法会尝试调用 `__sklearn_tags__`，但找不到该方法，从而抛出 `AttributeError`。 - 此外，如果没有实现 `fit`、`predict` 和 `score` 方法，`GridSearchCV` 也无法正常工作。 --- ###