Scikit-learn 基础

429

Scikit-learn 介绍

Scikit-learn 是开源的 Python 库,通过统一的界面实现机器学习、预处理、交叉验证及可视化算法。

scikit-learn

scikit-learn 网站:scikit-learn.org

Python 中的机器学习

  • 简单有效的数据挖掘和数据分析工具
  • 可供所有人访问,并可在各种环境中重复使用
  • 基于 NumPy,SciPy 和 matplotlib 构建
  • 开源,商业上可用 - BSD 许可证

ml_map

分类

确定对象属于哪个类别。

应用:垃圾邮件检测,图像识别。

算法: SVM,最近邻居,随机森林,......

回归

预测与对象关联的连续值属性。

应用:药物反应,股票价格。

算法: SVR,岭回归,套索,......

聚类

将类似对象自动分组到集合中。

应用:客户细分,分组实验结果

算法: k-Means,谱聚类,均值漂移,......

降维

减少要考虑的随机变量的数量。

应用:可视化,提高效率

算法: PCA,特征选择,非负矩阵分解。

模型选择

比较,验证和选择参数和模型。

目标:通过参数调整提高准确性

模块: 网格搜索,交叉验证,指标。

预处理

特征提取和规范化。

应用程序:转换输入数据(如文本)以与机器学习算法一起使用。 模块: 预处理,特征提取。

Scikit-learn 机器学习步骤

# 导入 sklearn
from sklearn import neighbors, datasets, preprocessing
from sklearn.model_selection import train_test_split
from sklearn.metrics import accuracy_score

# 加载数据
iris = datasets.load_iris()

# 划分训练集与测试集
X, y = iris.data[:, :2], iris.target
X_train, X_test, y_train, y_test = train_test_split(X, y, random_state=33)

# 数据预处理
scaler = preprocessing.StandardScaler().fit(X_train)
X_train = scaler.transform(X_train)
X_test = scaler.transform(X_test)

# 创建模型
knn = neighbors.KNeighborsClassifier(n_neighbors=5)
# 模型拟合
knn.fit(X_train, y_train)

# 预测
y_pred = knn.predict(X_test)
# 评估
accuracy_score(y_test, y_pred)

导入常用库

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt

加载数据

Scikit-learn 处理的数据是存储为 NumPy 数组或 SciPy 稀疏矩阵的数字,还支持 Pandas 数据框等可转换为数字数组的其它数据类型。

X = np.random.random((11, 5))
y = np.array(['M', 'M', 'F', 'F', 'M', 'F', 'M', 'M', 'F', 'F', 'F'])
X[X < 0.7] = 0

划分训练集与测试集

from sklearn.model_selection import train_test_split
X_train, X_test, y_train, y_test = train_test_split(X, y, random_state=0)

数据预处理

标准化

from sklearn.preprocessing import StandardScaler
scaler = StandardScaler().fit(X_train)
standardized_X = scaler.transform(X_train)
standardized_X_test = scaler.transform(X_test)

归一化

from sklearn.preprocessing import Normalizer
scaler = Normalizer().fit(X_train)
normalized_X = scaler.transform(X_train)
normalized_X_test = scaler.transform(X_test)

二值化

from sklearn.preprocessing import Binarizer
binarizer = Binarizer(threshold=0.0).fit(X)
binary_X = binarizer.transform(X)

编码分类特征

from sklearn.preprocessing import LabelEncoder
enc = LabelEncoder()
y = enc.fit_transform(y)

输入缺失值

from sklearn.preprocessing import Imputer
imp = Imputer(missing_values=0, strategy='mean', axis=0)
imp.fit_transform(X_train)

生成多项式特征

from sklearn.preprocessing import PolynomialFeatures
poly = PolynomialFeatures(5)
poly.fit_transform(X)

创建模型估计器

监督学习

# 线性回归
from sklearn.linear_model import LinearRegression
lr = LinearRegression(normalize=True)
# 支持向量机(SVM)
from sklearn.svm import SVC
svc = SVC(kernel='linear')
# 朴素贝叶斯
from sklearn.naive_bayes import GaussianNB
gnb = GaussianNB()
# KNN
from sklearn import neighbors
knn = neighbors.KNeighborsClassifier(n_neighbors=5)

无监督学习

# 主成分分析(PCA)
from sklearn.cluster import KMeans
from sklearn.decomposition import PCA
pca = PCA(n_components=0.95)
# K Means
k_means = KMeans(n_clusters=3, random_state=0)

拟合数据

监督学习

lr.fit(X, y)
knn.fit(X_train, y_train)
svc.fit(X_train, y_train)

无监督学习

k_means.fit(X_train)
pca_model = pca.fit_transform(X_train)

预测

监督学习

# 预测标签
y_pred = svc.predict(np.random.random((2,5)))
# 预测标签
y_pred = lr.predict(X_test)
# 评估标签概率
y_pred = knn.predict_proba(X_test)

无监督学习

y_pred = k_means.predict(X_test)

评估模型性能

分类指标

# 准确率
knn.score(X_test, y_test)
from sklearn.metrics import accuracy_score
accuracy_score(y_test, y_pred)
# 分类预估评价函数
from sklearn.metrics import classification_report
print(classification_report(y_test, y_pred))
# 混淆矩阵
from sklearn.metrics import confusion_matrix
print(confusion_matrix(y_test, y_pred))

回归指标

# 平均绝对误差
from sklearn.metrics import mean_absolute_error
y_true = [3, -0.5, 2]
mean_absolute_error(y_true, y_pred)
# 均方误差
from sklearn.metrics import mean_squared_error
mean_squared_error(y_test, y_pred)
# R2 评分
from sklearn.metrics import r2_score
r2_score(y_true, y_pred)

群集指标

# 调整兰德系数
from sklearn.metrics import adjusted_rand_score
adjusted_rand_score(y_true, y_pred)
# 同质性
from sklearn.metrics import homogeneity_score
homogeneity_score(y_true, y_pred)
# V-measure
from sklearn.metrics import v_measure_score
metrics.v_measure_score(y_true, y_pred)

交叉验证

from sklearn.cross_validation import cross_val_score
print(cross_val_score(knn, X_train, y_train, cv=4))
print(cross_val_score(lr, X, y, cv=2))

模型调整

网格搜索

from sklearn.grid search import GridSearchcV
params = {"n neighbors": np.arange(1, 3),
          "metric": ["euclidean", "cityblock"]}
grid = GridSearchCV(estimator=knn,
                    param_grid-params)
grid.fit(X_train, y_train)
print(grid.best score)
print(grid.best_estimator_.n_neighbors)

随机参数优化

from sklearn.grid_search import RandomizedSearchCV
params = {"n_neighbors": range(1, 5),
          "weights": ["uniform", "distance"]}
rsearch = RandomizedSearchCV(estimator=knn,
                             rsearch.fit(X_train, y_train) random_state=5)
print(rsearch.best_score_)