用Python实现：通过读取训练集、测试集、验证集三个文件，来构建由bp神经网络、朴素贝叶斯、extra-tree、逻辑回归构成的投票式集成学习模型

首先，需要导入所需的库：

import pandas as pd
from sklearn.preprocessing import LabelEncoder
from sklearn.neural_network import MLPClassifier
from sklearn.naive_bayes import GaussianNB
from sklearn.ensemble import ExtraTreesClassifier
from sklearn.linear_model import LogisticRegression
from sklearn.ensemble import VotingClassifier

然后，读取训练集、测试集、验证集数据：

train_data = pd.read_csv('train.csv')
test_data = pd.read_csv('test.csv')
valid_data = pd.read_csv('valid.csv')

接下来，对数据进行预处理，包括将标签进行编码、将特征和标签分开等：

le = LabelEncoder()
train_y = le.fit_transform(train_data['label'])
train_X = train_data.drop('label', axis=1)
test_y = le.transform(test_data['label'])
test_X = test_data.drop('label', axis=1)
valid_y = le.transform(valid_data['label'])
valid_X = valid_data.drop('label', axis=1)

然后，构建四个模型对象：

bp = MLPClassifier(hidden_layer_sizes=(100,), activation='relu', solver='adam', max_iter=500)
nb = GaussianNB()
et = ExtraTreesClassifier(n_estimators=100, random_state=42)
lr = LogisticRegression(random_state=42)

最后，构建投票式集成学习模型，并进行训练和预测：

ensemble = VotingClassifier(estimators=[('bp', bp), ('nb', nb), ('et', et), ('lr', lr)])
ensemble.fit(train_X, train_y)
print('Accuracy on training set:', ensemble.score(train_X, train_y))
print('Accuracy on validation set:', ensemble.score(valid_X, valid_y))
print('Accuracy on test set:', ensemble.score(test_X, test_y))

完整代码如下：

import pandas as pd
from sklearn.preprocessing import LabelEncoder
from sklearn.neural_network import MLPClassifier
from sklearn.naive_bayes import GaussianNB
from sklearn.ensemble import ExtraTreesClassifier
from sklearn.linear_model import LogisticRegression
from sklearn.ensemble import VotingClassifier

train_data = pd.read_csv('train.csv')
test_data = pd.read_csv('test.csv')
valid_data = pd.read_csv('valid.csv')

le = LabelEncoder()
train_y = le.fit_transform(train_data['label'])
train_X = train_data.drop('label', axis=1)
test_y = le.transform(test_data['label'])
test_X = test_data.drop('label', axis=1)
valid_y = le.transform(valid_data['label'])
valid_X = valid_data.drop('label', axis=1)

bp = MLPClassifier(hidden_layer_sizes=(100,), activation='relu', solver='adam', max_iter=500)
nb = GaussianNB()
et = ExtraTreesClassifier(n_estimators=100, random_state=42)
lr = LogisticRegression(random_state=42)

ensemble = VotingClassifier(estimators=[('bp', bp), ('nb', nb), ('et', et), ('lr', lr)])
ensemble.fit(train_X, train_y)

print('Accuracy on training set:', ensemble.score(train_X, train_y))
print('Accuracy on validation set:', ensemble.score(valid_X, valid_y))
print('Accuracy on test set:', ensemble.score(test_X, test_y))