smote算法python代码

下面是使用Python实现SMOTE算法的示例代码：

from collections import Counter
import numpy as np
from sklearn.neighbors import NearestNeighbors

def SMOTE(T, N, k):
    """
    T: 原始数据集，N: 生成新数据的个数，k: 选择最近邻的数目
    """
    n_samples, n_features = T.shape
    if N < 100:
        N = 100
    N = int(N / 100) * n_samples
    neighbors = NearestNeighbors(n_neighbors=k).fit(T)
    new_samples = np.zeros((N, n_features))
    for i in range(N):
        j = np.random.randint(0, n_samples)
        nn = np.random.choice(neighbors.kneighbors(T[j].reshape(1, -1), return_distance=False)[0])
        dif = T[nn] - T[j]
        gap = np.random.rand()
        new_samples[i] = T[j] + gap * dif
    return new_samples

def balance_data(X, y, N, k):
    """
    X: 原始数据集，y: 数据集对应的标签，N: 生成新数据的个数，k: 选择最近邻的数目
    """
    counter = Counter(y)
    max_num = max(counter.values())
    classes = list(counter.keys())
    X_resampled = X.copy()
    y_resampled = y.copy()
    for target_class in classes:
        target_num = max_num - counter[target_class]
        if target_num > 0:
            target_X = X[y==target_class]
            new_samples = SMOTE(target_X, target_num, k)
            new_labels = np.array([target_class] * target_num)
            X_resampled = np.concatenate((X_resampled, new_samples), axis=0)
            y_resampled = np.concatenate((y_resampled, new_labels), axis=0)
    return X_resampled, y_resampled

使用示例：

from sklearn.datasets import make_classification
from sklearn.model_selection import train_test_split
from sklearn.svm import SVC

# 生成一个二分类的数据集
X, y = make_classification(n_samples=1000, n_features=10, n_classes=2, weights=[0.9, 0.1], random_state=42)

# 将原始数据集分为训练集和测试集
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=42)

# 使用SMOTE算法平衡训练集
X_train_resampled, y_train_resampled = balance_data(X_train, y_train, N=5000, k=5)

# 训练一个支持向量机模型
clf = SVC(kernel='linear', random_state=42)
clf.fit(X_train_resampled, y_train_resampled)

# 测试模型
print(clf.score(X_test, y_test))