9.30
1)从 scikit-learn 库中加载 iris 数据集,使用留出法留出 1/3 的样本作为测试集(注
意同分布取样);
(2)使用训练集训练 BP 神经网络分类算法;
(3)使用五折交叉验证对模型性能(准确度、精度、召回率和 F1 值)进行评估和选
择;
(4)使用测试集,测试模型的性能,对测试结果进行分析
import numpy as np
from sklearn.datasets import load_iris
from sklearn.model_selection import train_test_split, cross_val_score, KFold
from sklearn.neural_network import MLPClassifier
from sklearn.metrics import accuracy_score, precision_score, recall_score, f1_score
# 加载iris数据集
iris = load_iris()
X = iris.data
y = iris.target
# 使用留出法划分训练集和测试集
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=1/3, random_state=42, stratify=y)
# 训练BP神经网络模型
bpnn = MLPClassifier(hidden_layer_sizes=(10,), max_iter=1000, random_state=42)
bpnn.fit(X_train, y_train)
# 五折交叉验证评估模型性能
kf = KFold(n_splits=5, shuffle=True, random_state=42)
cv_scores = cross_val_score(bpnn, X_train, y_train, cv=kf, scoring='accuracy')
precision_scores = cross_val_score(bpnn, X_train, y_train, cv=kf, scoring='precision_macro')
recall_scores = cross_val_score(bpnn, X_train, y_train, cv=kf, scoring='recall_macro')
f1_scores = cross_val_score(bpnn, X_train, y_train, cv=kf, scoring='f1_macro')
print("五折交叉验证准确度:", cv_scores.mean())
print("五折交叉验证精度:", precision_scores.mean())
print("五折交叉验证召回率:", recall_scores.mean())
print("五折交叉验证F1值:", f1_scores.mean())
# 使用测试集测试模型性能
y_pred = bpnn.predict(X_test)
accuracy = accuracy_score(y_test, y_pred)
precision = precision_score(y_test, y_pred, average='macro')
recall = recall_score(y_test, y_pred, average='macro')
f1 = f1_score(y_test, y_pred, average='macro')
print("测试集准确度:", accuracy)
print("测试集精度:", precision)
print("测试集召回率:", recall)
print("测试集F1值:", f1)
# 测试结果分析
# 比较五折交叉验证和测试集的评估指标
if np.abs(accuracy - cv_scores.mean()) < 0.1:
print("模型的泛化能力较好,在训练集交叉验证和测试集上的准确度较为接近。")
else:
print("模型可能存在过拟合或欠拟合问题,训练集交叉验证和测试集上的准确度差异较大。")
# 分析各类别样本的预测情况
from collections import Counter
class_pred_count = Counter(y_pred)
class_true_count = Counter(y_test)
for class_label in range(3):
class_accuracy = accuracy_score(y_test[y_test == class_label], y_pred[y_test == class_label])
print(f"类别{class_label}的准确度: {class_accuracy}")
if class_pred_count[class_label] == 0:
print(f"类别{class_label}在预测结果中未出现,可能存在对该类别的学习不足问题。")
elif class_true_count[class_label] == 0:
print(f"测试集中没有类别{class_label}的样本,可能数据划分存在问题。")


浙公网安备 33010602011771号