机器学习之学习曲线

第一部分：基本含义

学习曲线（Learning Curve）是机器学习中用于评估模型表现的一个重要工具，它可以帮助我们直观地理解模型的学习过程，并诊断模型的状态，例如是否存在欠拟合（underfitting）或过拟合（overfitting）。学习曲线通常显示训练误差和验证（测试）误差随着训练集大小的变化情况。

在正常拟合的情况下，训练集和测试集的损失值随之训练样本增加的变化情况大致如下

第二部分：代码实现

（1）导包

from sklearn.linear_model import LinearRegression
from sklearn.preprocessing import PolynomialFeatures
import numpy as np
import matplotlib.pyplot as plt
from sklearn.model_selection import train_test_split

（2）创建数据集

# 第二部分：获取数据集
np.random.seed(233)
x = np.random.uniform(-4, 2, size=(100))
y = x ** 2 + 4 * x + 3 + 2 * np.random.randn(100)
x = x.reshape(-1, 1)

#plt.scatter(x, y)
#plt.show()

（3）划分数据集

# 第三部分：划分数据集
x_train, x_test, y_train, y_test = train_test_split(x, y, train_size=0.7, random_state=233)

（4）开始画出学习曲线

#第四部分：开始绘制学习曲线
import matplotlib.pyplot as plt

plt.rcParams["figure.figsize"] = (12, 8)

degrees = [1, 2, 5, 20]
for i, degree in enumerate(degrees):
    polynomial_features = PolynomialFeatures(degree=degree)

    X_poly_train = polynomial_features.fit_transform(x_train.reshape(-1, 1))
    X_poly_test = polynomial_features.fit_transform(x_test.reshape(-1, 1))

    train_error, test_error = [], []
    for k in range(1,len(x_train)):
        linear_regression = LinearRegression()
        linear_regression.fit(X_poly_train[:k + 1], y_train[:k + 1])

        train_error.append(linear_regression.score(X_poly_train[:k + 1], y_train[:k + 1]))
        test_error.append(linear_regression.score(X_poly_test, y_test))

    plt.subplot(2, 2, i + 1)
    plt.title("Degree: {0}".format(degree))
    plt.ylim(-1, 1)
    plt.plot([k + 1 for k in range(1,len(x_train))], train_error, color="red", label='train')
    plt.plot([k + 1 for k in range(1,len(x_train))], test_error, color="blue", label='test')
plt.tight_layout()  # 自动调整子图间距
plt.show()

（5）完整pycharm代码实现

from sklearn.linear_model import LinearRegression
from sklearn.preprocessing import PolynomialFeatures
import numpy as np
import matplotlib.pyplot as plt
from sklearn.model_selection import train_test_split

# 第二部分：获取数据集
np.random.seed(233)
x = np.random.uniform(-4, 2, size=(100))
y = x ** 2 + 4 * x + 3 + 2 * np.random.randn(100)
x = x.reshape(-1, 1)

plt.scatter(x, y)

# 第三部分：划分数据集
x_train, x_test, y_train, y_test = train_test_split(x, y, train_size=0.7, random_state=233)

#第四部分：开始绘制学习曲线
import matplotlib.pyplot as plt

plt.rcParams["figure.figsize"] = (12, 8)

degrees = [1, 2, 5, 20]
for i, degree in enumerate(degrees):
    polynomial_features = PolynomialFeatures(degree=degree)

    X_poly_train = polynomial_features.fit_transform(x_train.reshape(-1, 1))
    X_poly_test = polynomial_features.fit_transform(x_test.reshape(-1, 1))

    train_error, test_error = [], []
    for k in range(1,len(x_train)):
        linear_regression = LinearRegression()
        linear_regression.fit(X_poly_train[:k + 1], y_train[:k + 1])

        train_error.append(linear_regression.score(X_poly_train[:k + 1], y_train[:k + 1]))
        test_error.append(linear_regression.score(X_poly_test, y_test))

    plt.subplot(2, 2, i + 1)
    plt.title("Degree: {0}".format(degree))
    plt.ylim(-1, 1)
    plt.plot([k + 1 for k in range(1,len(x_train))], train_error, color="red", label='train')
    plt.plot([k + 1 for k in range(1,len(x_train))], test_error, color="blue", label='test')
plt.tight_layout()  # 自动调整子图间距
plt.show()

从上述图中的运行结果来看：

Degree=1的时候，蓝色线（训练集train）不是从小到大进行的，degree=5和degree=20的时候波动很大，只有degree=2的时候效果好，刚好也符合我们的二次函数的数据集本身的特性（y = x ** 2 + 4 * x + 3 + 2 * np.random.randn(100)）