当前位置：首页 > news >正文

Scikit-learn (sklearn) 库详细介绍

news 2025/8/17 23:29:40

Scikit-learn (简称 sklearn) 是 Python 中最流行的机器学习库之一，它建立在 NumPy、SciPy 和 Matplotlib 之上，提供了简单高效的数据挖掘和数据分析工具。

1. 主要特点

简单易用：一致的 API 设计，学习曲线平缓
功能全面：覆盖了从数据预处理到模型评估的完整机器学习流程
高效性能：基于 NumPy 和 SciPy 实现，运算效率高
开源免费：BSD 许可证下的开源项目
文档完善：提供详细的文档和丰富的示例

2. 主要功能模块

2.1 数据预处理 (sklearn.preprocessing)

from sklearn import preprocessing# 标准化
scaler = preprocessing.StandardScaler().fit(X_train)
X_train_scaled = scaler.transform(X_train)# 归一化
normalizer = preprocessing.MinMaxScaler().fit(X_train)
X_train_normalized = normalizer.transform(X_train)# 编码分类特征
encoder = preprocessing.OneHotEncoder().fit(X_categorical)
X_encoded = encoder.transform(X_categorical)

2.2 特征选择 (sklearn.feature_selection)

from sklearn.feature_selection import SelectKBest, chi2# 选择K个最好的特征
selector = SelectKBest(chi2, k=10)
X_new = selector.fit_transform(X, y)

2.3 模型选择 (sklearn.model_selection)

from sklearn.model_selection import train_test_split, cross_val_score, GridSearchCV# 数据集划分
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2)# 交叉验证
scores = cross_val_score(model, X, y, cv=5)# 网格搜索调参
param_grid = {'C': [0.1, 1, 10], 'gamma': [0.01, 0.1]}
grid_search = GridSearchCV(SVC(), param_grid, cv=5)
grid_search.fit(X_train, y_train)

2.4 监督学习算法

线性模型 (sklearn.linear_model)

from sklearn.linear_model import LinearRegression, LogisticRegression# 线性回归
lr = LinearRegression().fit(X_train, y_train)# 逻辑回归
logreg = LogisticRegression().fit(X_train, y_train)

支持向量机 (sklearn.svm)

from sklearn.svm import SVC, SVR# 分类
svc = SVC(kernel='rbf', C=1.0).fit(X_train, y_train)# 回归
svr = SVR(kernel='linear').fit(X_train, y_train)

决策树 (sklearn.tree)

from sklearn.tree import DecisionTreeClassifier, DecisionTreeRegressor# 分类树
dt_clf = DecisionTreeClassifier(max_depth=5).fit(X_train, y_train)# 回归树
dt_reg = DecisionTreeRegressor().fit(X_train, y_train)

集成方法 (sklearn.ensemble)

from sklearn.ensemble import RandomForestClassifier, GradientBoostingRegressor# 随机森林
rf = RandomForestClassifier(n_estimators=100).fit(X_train, y_train)# 梯度提升树
gbr = GradientBoostingRegressor().fit(X_train, y_train)

2.5 无监督学习算法

聚类 (sklearn.cluster)

from sklearn.cluster import KMeans, DBSCAN# K均值聚类
kmeans = KMeans(n_clusters=3).fit(X)# DBSCAN聚类
dbscan = DBSCAN(eps=0.5).fit(X)

降维 (sklearn.decomposition)

from sklearn.decomposition import PCA, TruncatedSVD# 主成分分析
pca = PCA(n_components=2).fit(X)
X_pca = pca.transform(X)# 截断SVD
svd = TruncatedSVD(n_components=100).fit(X)

2.6 模型评估 (sklearn.metrics)

from sklearn.metrics import accuracy_score, confusion_matrix, classification_report# 分类评估
y_pred = model.predict(X_test)
print("Accuracy:", accuracy_score(y_test, y_pred))
print("Confusion Matrix:\n", confusion_matrix(y_test, y_pred))
print("Classification Report:\n", classification_report(y_test, y_pred))# 回归评估
from sklearn.metrics import mean_squared_error, r2_score
print("MSE:", mean_squared_error(y_test, y_pred))
print("R2 Score:", r2_score(y_test, y_pred))

3. 工作流程示例

# 1. 导入必要的库和模块
from sklearn.datasets import load_iris
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler
from sklearn.ensemble import RandomForestClassifier
from sklearn.metrics import classification_report# 2. 加载数据集
iris = load_iris()
X, y = iris.data, iris.target# 3. 数据预处理
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2)
scaler = StandardScaler().fit(X_train)
X_train = scaler.transform(X_train)
X_test = scaler.transform(X_test)# 4. 训练模型
model = RandomForestClassifier(n_estimators=100)
model.fit(X_train, y_train)# 5. 评估模型
y_pred = model.predict(X_test)
print(classification_report(y_test, y_pred))

4. 实用技巧

1.管道(Pipeline)使用：将多个处理步骤组合在一起

from sklearn.pipeline import Pipeline
pipe = Pipeline([('scaler', StandardScaler()),('selector', SelectKBest(k=10)),('classifier', RandomForestClassifier())
])

2.特征联合：组合不同类型的特征

from sklearn.pipeline import FeatureUnion
from sklearn.decomposition import PCA
from sklearn.feature_selection import SelectKBestcombined_features = FeatureUnion([("pca", PCA(n_components=3)),("univ_select", SelectKBest(k=1))
])

3.自定义转换器：创建自己的预处理步骤

from sklearn.base import BaseEstimator, TransformerMixinclass MyTransformer(BaseEstimator, TransformerMixin):def fit(self, X, y=None):return selfdef transform(self, X):# 实现你的转换逻辑return X_transformed