From a3730c4840e71a4ae5ce2a1b88aab041bfd1e5cb Mon Sep 17 00:00:00 2001
From: mogu-king <1547976000@qq.com>
Date: Sun, 3 Nov 2024 04:33:11 +0800
Subject: [PATCH] =?UTF-8?q?=E6=9B=B4=E6=96=B0=20function.py?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 function.py | 43 +++++++++++++++++++++++++++++++++++--------
 1 file changed, 35 insertions(+), 8 deletions(-)

diff --git a/function.py b/function.py
index bc95f6b..8182d90 100644
--- a/function.py
+++ b/function.py
@@ -1,5 +1,5 @@
 import numpy as np
-
+from sklearn.tree import DecisionTreeClassifier
 """
 实验说明:
 - 自行编写实现数据加载与预测病马死亡的对率回归算法。
@@ -16,6 +16,7 @@ import numpy as np
 id = "202210423208"
 name = "刘义天"
 
+
 # 加载数据并添加非线性特征
 def load_data(filepath):
     """
@@ -78,20 +79,46 @@ def train_logreg(X, y, learning_rate=0.01, num_iterations=1000, regularization_s
     return weights
 
 
-# 使用 0.4 阈值的对率回归预测函数
-def predict_logreg(inX, weights):
+# 决策树训练函数
+def train_decision_tree(X, y):
     """
-    对率回归预测函数。
+    训练简单的决策树模型。
+
+    Args:
+    - X (numpy array): 特征数据
+    - y (numpy array): 标签数据
+
+    Returns:
+    - clf (DecisionTreeClassifier): 训练好的决策树模型
+    """
+    clf = DecisionTreeClassifier(max_depth=5, random_state=42)
+    clf.fit(X, y)
+    return clf
+
+
+# 融合预测函数
+def predict_fusion(inX, weights, clf):
+    """
+    融合对率回归和决策树的预测结果。
 
     Args:
     - inX (numpy array): 输入待分类的样本。
-    - weights (numpy array): 训练得到的权重。
+    - weights (numpy array): 对率回归的权重。
+    - clf (DecisionTreeClassifier): 训练好的决策树模型。
 
     Returns:
-    - int: 预测的分类标签。
+    - int: 最终预测的分类标签。
     """
-    probability = 1 / (1 + np.exp(-np.dot(inX, weights)))
-    return 1 if probability >= 0.4 else 0
+    # 对率回归预测
+    probability_logreg = 1 / (1 + np.exp(-np.dot(inX, weights)))
+    prediction_logreg = 1 if probability_logreg >= 0.4 else 0
+
+    # 决策树预测
+    prediction_tree = clf.predict(inX.reshape(1, -1))[0]
+    
+    # 融合投票
+    final_prediction = 1 if (prediction_logreg + prediction_tree) >= 1 else 0
+    return final_prediction
 
 if __name__ == "__main__":
     # 以下代码可用于测试，可自行随意修改