多模态数据决策层融合

# 数据准备

# 特征集1
X1 = ...
y1 = ...

# 特征集2
X2 = ...
y2 = ...

# 特征集3
X3 = ...
y3 = ...

# 单模态分类

from sklearn.model_selection import train_test_split
from sklearn.tree import DecisionTreeClassifier
from sklearn.metrics import accuracy_score

# 训练模型1
X_train1, X_test1, y_train1, y_test1 = train_test_split(X1, y1, test_size=0.2)
clf1 = DecisionTreeClassifier()
clf1.fit(X_train1, y_train1)
y_pred1 = clf1.predict(X_test1)
accuracy1 = accuracy_score(y_test1, y_pred1)

# 训练模型2
X_train2, X_test2, y_train2, y_test2 = train_test_split(X2, y2, test_size=0.2)
clf2 = DecisionTreeClassifier()
clf2.fit(X_train2, y_train2)
y_pred2 = clf2.predict(X_test2)
accuracy2 = accuracy_score(y_test2, y_pred2)

# 训练模型3
X_train3, X_test3, y_train3, y_test3 = train_test_split(X3, y3, test_size=0.2)
clf3 = DecisionTreeClassifier()
clf3.fit(X_train3, y_train3)
y_pred3 = clf3.predict(X_test3)
accuracy3 = accuracy_score(y_test3, y_pred3)

# 特征融合

import numpy as np

# 加权平均融合
y_pred_avg = (accuracy1 * y_pred1 + accuracy2 * y_pred2 + accuracy3 * y_pred3) / (accuracy1 + accuracy2 + accuracy3)

# 投票融合
y_pred_vote = np.argmax(np.vstack((y_pred1, y_pred2, y_pred3)), axis=0)

# 堆叠融合
X_train_stack = np.vstack((y_pred1, y_pred2, y_pred3)).T
X_test_stack = np.vstack((clf1.predict(X_test1), clf2.predict(X_test2), clf3.predict(X_test3))).T
meta_clf = DecisionTreeClassifier()
meta_clf.fit(X_train_stack, y_test1)  # 使用任意一个真实标签作为目标变量
y_pred_stack = meta_clf.predict(X_test_stack)

# 性能评估
accuracy_avg = accuracy_score(y_test1, y_pred_avg)
accuracy_vote = accuracy_score(y_test1, y_pred_vote)
accuracy_stack = accuracy_score(y_test1, y_pred_stack)

print("加权平均融合准确率:", accuracy_avg)
print("投票融合准确率:", accuracy_vote)
print("堆叠融合准确率:", accuracy_stack)