我正在尝试在神经网络中使用sklearn和TensorFlow包实现KFold验证。
我的代码如下所示。
def training(self): n_split = 3 instances = self.instance labels = self.labels for train_index, test_index in KFold(n_split).split(instances): x_train, x_test = instances[train_index], instances[test_index] y_train, y_test = labels[train_index], labels[test_index] model = self.mlp_model() model.fit(x_train, y_train, epochs=20) print('Model Evaluation', model.evaluate(x_test, y_test))遗憾的是，我遇到了一个错误
raise KeyError(f"None of [{key}] are in the [{axis_name}]")KeyError: "None of [Int64Index([160, 161, 162, 163, 164, 165, 166, 167, 168, 169,\n ...\n 468, 469, 470, 471, 472, 473, 474, 475, 476, 477],\n dtype='int64', length=318)] are in the [columns]"我的instances的形状是(478,15)，labels的形状是(478,)。
我的instances数据集看起来像这样
| | SibSp | Parch | Female | Male | C | Q | S | 1 | 2 | 3 | Master ||-----|-------|-------|--------|------|---|---|---|---|---|---|--------|| 374 | 0 | 0 | 0 | 1 | 0 | 0 | 1 | 0 | 0 | 1 | 1 || 579 | 0 | 0 | 0 | 1 | 0 | 0 | 1 | 0 | 0 | 1 | 0 || 784 | 0 | 0 | 0 | 1 | 1 | 0 | 1 | 0 | 0 | 1 | 0 |labels看起来像这样
| | Age ||-----|-----|| 127 | 24 || 262 | 52 || 615 | 24 |问题的可能原因是什么？可能的解决方案是什么？期待您的回复。提前感谢您。
回答：
这是一个使用KFold交叉验证在TensorFlow的CIFAR10数据集上的代码示例。
获取数据
import tensorflow as tfimport numpy as np (input, target), (_, _) = tf.keras.datasets.cifar10.load_data()# 将数字解析为浮点数input = input.astype('float32') / 255target = tf.keras.utils.to_categorical(target , num_classes=10)print(input.shape, target.shape)# (50000, 32, 32, 3) (50000, 10)模型
def my_model(): return tf.keras.Sequential( [ tf.keras.Input(shape=(32, 32, 3)), tf.keras.layers.Conv2D(16, 3, activation="relu"), tf.keras.layers.Conv2D(32, 3, activation="relu"), tf.keras.layers.Conv2D(64, 3, activation="relu"), tf.keras.layers.Conv2D(128, 3, activation="relu"), tf.keras.layers.Conv2D(256, 3, activation="relu"), tf.keras.layers.GlobalAveragePooling2D(), tf.keras.layers.Dense(10, activation='softmax') ] )我们将在KFold循环中调用这个模型。
K-Fold训练
from sklearn.model_selection import KFoldimport numpy as npfor kfold, (train, test) in enumerate(KFold(n_splits=3, shuffle=True).split(input, target)): # 清除会话 tf.keras.backend.clear_session() # 调用并编译模型 seq_model = my_model() seq_model.compile( loss = tf.keras.losses.CategoricalCrossentropy(), metrics = tf.keras.metrics.CategoricalAccuracy(), optimizer = tf.keras.optimizers.Adam()) print('训练集') print(input[train].shape) print(target[train].shape) print('测试集') print(input[test].shape) print(target[test].shape) # 运行模型 seq_model.fit(input[train], target[train], batch_size=128, epochs=2, validation_data=(input[test], target[test])) seq_model.save_weights(f'wg_{kfold}.h5')日志
训练集(33333, 32, 32, 3)(33333, 10)测试集(16667, 32, 32, 3)(16667, 10)Epoch 1/211s 41ms/step - loss: 1.9961 - categorical_accuracy: 0.2363 - val_loss: 1.6851 - val_categorical_accuracy: 0.3435Epoch 2/210s 37ms/step - loss: 1.6322 - categorical_accuracy: 0.3836 - val_loss: 1.5780 - val_categorical_accuracy: 0.4193训练集(33333, 32, 32, 3)(33333, 10)测试集(16667, 32, 32, 3)(16667, 10)Epoch 1/211s 39ms/step - loss: 2.0254 - categorical_accuracy: 0.2197 - val_loss: 1.6799 - val_categorical_accuracy: 0.3601Epoch 2/210s 37ms/step - loss: 1.6687 - categorical_accuracy: 0.3739 - val_loss: 1.5222 - val_categorical_accuracy: 0.4362训练集(33334, 32, 32, 3)(33334, 10)测试集(16666, 32, 32, 3)(16666, 10)Epoch 1/211s 41ms/step - loss: 2.0170 - categorical_accuracy: 0.2212 - val_loss: 1.7452 - val_categorical_accuracy: 0.3134Epoch 2/210s 37ms/step - loss: 1.7110 - categorical_accuracy: 0.3363 - val_loss: 1.5928 - val_categorical_accuracy: 0.4164更新
另一个使用pandas数据框的方法。
# 加载df = pd.read_csv('train.csv')# [可选]: 打乱数据df = df.sample(frac=1).reset_index(drop=True)# 折叠分割 kfold = KFold(n_splits=3, shuffle=True)for each_fold, (trn_idx, val_idx) in enumerate(.split(np.arange(df.shape[0]), df.target.values)): # 获取折叠数据 train_labels = df.iloc[trn_idx].reset_index(drop=True) val_labels = df.iloc[val_idx].reset_index(drop=True) # train_labels: 训练对（数据 + 标签） # 做一些事情让我们再给出一个多标签数据的示例。这次我们将首先创建一个折叠列。请看下面的示例。
# !pip install iterative-stratificationfrom iterstrat.ml_stratifiers import MultilabelStratifiedKFold# 加载数据 df = pd.read_csv('train.csv')# 添加一个新列，目前所有值设置为-1# 我们稍后会更新它 df.loc[:, 'kfold'] = -1 # [可选]: 打乱 df = df.sample(frac=1).reset_index(drop=True)# 由于我们现在处理的是多标签情况， # 目标值会有不止一列 # 所以我们只获取目标值 # 我们将删除 id / image_id / image_name / blabla.. 列target = df.drop('id', axis-1).values mskf = MultilabelStratifiedKFold(n_splits=5)for each_kfold, (trn_idx, val_idx) in enumerate(mskf.split(df, target): df.loc[val_idx, 'kfold] = each_fold # 更新 `kfold` 列 之后，n_split = 5 将出现在 ‘kfold’ 列中，也可以使用 df.head() 进行检查。接下来，我们可以做如下操作
def program(fold): # 获取折叠数据 train_labels = df[df.kfold != fold].reset_index(drop=True) val_labels = df[df.kfold == fold].reset_index(drop=True) # train_labels: 训练对（数据 + 标签） # 做一些事情for i in range(n_split): program(fold=i) 