Scikit-学习标签编码，然后进行一次热编码，为训练和测试数据集产生不同的特征集。如何解决这个问题？

from sklearn.preprocessing import LabelEncoder, OneHotEncoder

# 假设训练数据集的标签为['A', 'B', 'C', 'A', 'B']
# 假设测试数据集的标签为['C', 'A', 'B']

# 标签编码
label_encoder = LabelEncoder()
train_labels = ['A', 'B', 'C', 'A', 'B']
test_labels = ['C', 'A', 'B']
encoded_train_labels = label_encoder.fit_transform(train_labels)
encoded_test_labels = label_encoder.transform(test_labels)

# 热编码
onehot_encoder = OneHotEncoder(sparse=False)
encoded_train_labels = encoded_train_labels.reshape(len(encoded_train_labels), 1)
encoded_test_labels = encoded_test_labels.reshape(len(encoded_test_labels), 1)
onehot_train_labels = onehot_encoder.fit_transform(encoded_train_labels)
onehot_test_labels = onehot_encoder.transform(encoded_test_labels)

# 打印结果
print("训练数据集的热编码标签：")
print(onehot_train_labels)
print("测试数据集的热编码标签：")
print(onehot_test_labels)