实现五折交叉验证进行模型训练

1、实验目的
熟悉Python 的基本操作，掌握对数据集的读写实现、对模型性能的评估实现的能力；
加深对训练集、测试集、N折交叉验证、模型评估标准的理解。
2、实验内容
（1）利用pandas库从本地读取iris数据集；
（2）从scikit-learn 库中直接加载iris 数据集；
（3）实现五折交叉验证进行模型训练；
（4）计算并输出模型的准确度、精度、召回率和F1值。
3、操作要点
（1）安装Python及pycharm（一种Python开发IDE），并熟悉Python基本操作；
（2）学习pandas库里存取文件的相关函数，以及scikit-learn库里数据集下载、交叉验
证、模型评估等相关操作；
（3）可能用的库有pandas，scikit-learn，numpy 等，需要提前下载pip；
（4）测试模型可使用随机森林rf_classifier=RandomForestClassifier(n_estimators=100)，
或其它分类器

实验代码：

"""
Iris数据集分类任务
实现五折交叉验证并计算模型评估指标
"""

import pandas as pd
import numpy as np
from sklearn.datasets import load_iris
from sklearn.model_selection import KFold, cross_val_predict
from sklearn.linear_model import LogisticRegression
from sklearn.preprocessing import StandardScaler
from sklearn.metrics import accuracy_score, precision_score, recall_score, f1_score
import warnings
import sys
import io

# 设置输出编码为UTF-8，解决Windows PowerShell中文乱码问题
if sys.platform == 'win32':
    sys.stdout = io.TextIOWrapper(sys.stdout.buffer, encoding='utf-8')
    sys.stderr = io.TextIOWrapper(sys.stderr.buffer, encoding='utf-8')

warnings.filterwarnings('ignore')

# # ==================== 步骤1：使用pandas从本地读取iris数据集 ====================
# print("=" * 70)
# print("步骤1：使用pandas从本地读取iris数据集")
# print("=" * 70)
#
# # 读取本地iris数据文件
# iris_local_path = 'iris/iris.data'
# iris_local = pd.read_csv(iris_local_path, header=None,
#                          names=['sepal_length', 'sepal_width', 'petal_length', 'petal_width', 'species'])
#
# print(f"本地数据集形状: {iris_local.shape}")
# print(f"前5行数据:")
# print(iris_local.head())
# print(f"\n类别分布:")
# print(iris_local['species'].value_counts())
# print()
#
# # 分离特征和标签
# X_local = iris_local.iloc[:, :-1].values
# y_local = iris_local.iloc[:, -1].values

# ==================== 步骤2：从scikit-learn库中直接加载iris数据集 ====================
print("=" * 70)
print("步骤2：从scikit-learn库中直接加载iris数据集")
print("=" * 70)

# 从sklearn加载iris数据集
iris_sklearn = load_iris()
X_sklearn = iris_sklearn.data
y_sklearn = iris_sklearn.target
feature_names = iris_sklearn.feature_names
target_names = iris_sklearn.target_names

print(f"Sklearn数据集形状: {X_sklearn.shape}")
print(f"特征名称: {feature_names}")
print(f"类别名称: {target_names}")
print(f"类别分布: {np.bincount(y_sklearn)}")
print()

# 将sklearn数据转换为DataFrame以便查看
iris_sklearn_df = pd.DataFrame(X_sklearn, columns=feature_names)
iris_sklearn_df['species'] = [target_names[i] for i in y_sklearn]
print("前5行数据:")
print(iris_sklearn_df.head())
print()

# ==================== 步骤3：实现五折交叉验证进行模型训练 ====================
print("=" * 70)
print("步骤3：实现五折交叉验证进行模型训练")
print("=" * 70)

# 选择使用sklearn加载的数据集（更标准）
X = X_sklearn
y = y_sklearn

# 数据标准化
scaler = StandardScaler()
X_scaled = scaler.fit_transform(X)

# 创建模型（使用逻辑回归作为示例）
model = LogisticRegression(random_state=42, max_iter=1000)

# 创建五折交叉验证
kfold = KFold(n_splits=5, shuffle=True, random_state=42)

# 存储每折的预测结果
all_y_true = []
all_y_pred = []

print("开始五折交叉验证...")
fold_num = 1
for train_idx, test_idx in kfold.split(X_scaled):
    print(f"\n第 {fold_num} 折:")
    print(f"  训练集大小: {len(train_idx)}, 测试集大小: {len(test_idx)}")
    
    # 划分训练集和测试集
    X_train, X_test = X_scaled[train_idx], X_scaled[test_idx]
    y_train, y_test = y[train_idx], y[test_idx]
    
    # 训练模型
    model.fit(X_train, y_train)
    
    # 预测
    y_pred = model.predict(X_test)
    
    # 保存预测结果
    all_y_true.extend(y_test)
    all_y_pred.extend(y_pred)
    
    # 计算当前折的指标
    acc = accuracy_score(y_test, y_pred)
    prec = precision_score(y_test, y_pred, average='weighted', zero_division=0)
    rec = recall_score(y_test, y_pred, average='weighted', zero_division=0)
    f1 = f1_score(y_test, y_pred, average='weighted', zero_division=0)
    
    print(f"  准确度: {acc:.4f}")
    print(f"  精度: {prec:.4f}")
    print(f"  召回率: {rec:.4f}")
    print(f"  F1值: {f1:.4f}")
    
    fold_num += 1

print("\n五折交叉验证完成！")
print()

# ==================== 步骤4：计算并输出模型的准确度、精度、召回率和F1值 ====================
print("=" * 70)
print("步骤4：计算并输出模型的评估指标（整体结果）")
print("=" * 70)

# 计算整体指标
accuracy = accuracy_score(all_y_true, all_y_pred)
precision = precision_score(all_y_true, all_y_pred, average='weighted', zero_division=0)
recall = recall_score(all_y_true, all_y_pred, average='weighted', zero_division=0)
f1 = f1_score(all_y_true, all_y_pred, average='weighted', zero_division=0)

print(f"\n整体评估指标（基于所有5折的预测结果）:")
print(f"  准确度 (Accuracy): {accuracy:.4f}")
print(f"  精度 (Precision): {precision:.4f}")
print(f"  召回率 (Recall): {recall:.4f}")
print(f"  F1值 (F1-Score): {f1:.4f}")
print()

# 使用cross_val_predict方法（另一种方式）
print("=" * 70)
print("使用cross_val_predict方法进行交叉验证（对比）")
print("=" * 70)

y_pred_cv = cross_val_predict(model, X_scaled, y, cv=5)

accuracy_cv = accuracy_score(y, y_pred_cv)
precision_cv = precision_score(y, y_pred_cv, average='weighted', zero_division=0)
recall_cv = recall_score(y, y_pred_cv, average='weighted', zero_division=0)
f1_cv = f1_score(y, y_pred_cv, average='weighted', zero_division=0)

print(f"\n评估指标（cross_val_predict方法）:")
print(f"  准确度 (Accuracy): {accuracy_cv:.4f}")
print(f"  精度 (Precision): {precision_cv:.4f}")
print(f"  召回率 (Recall): {recall_cv:.4f}")
print(f"  F1值 (F1-Score): {f1_cv:.4f}")
print()

# 按类别显示详细指标
print("=" * 70)
print("按类别显示详细指标")
print("=" * 70)

precision_per_class = precision_score(all_y_true, all_y_pred, average=None, zero_division=0)
recall_per_class = recall_score(all_y_true, all_y_pred, average=None, zero_division=0)
f1_per_class = f1_score(all_y_true, all_y_pred, average=None, zero_division=0)

print("\n各类别指标:")
print(f"{'类别':<20} {'精度':<12} {'召回率':<12} {'F1值':<12}")
print("-" * 60)
for i, class_name in enumerate(target_names):
    print(f"{class_name:<20} {precision_per_class[i]:<12.4f} {recall_per_class[i]:<12.4f} {f1_per_class[i]:<12.4f}")

print("\n" + "=" * 70)
print("任务完成！")
print("=" * 70)

实验结果得到的数据：