scikit-learn的基本用法(四)——数据归一化

文章作者：Tyan
博客：noahsnail.com | CSDN | 简书

本文主要介绍scikit-learn中的数据预处理之归一化。

Demo 1

import numpy as np
from sklearn import preprocessing


# 定义array
a = np.array([-10, 2.3, 13.7, 56, 108])
print a

# 对array进行归一化(normalization)
# scale进行的操作是按列减去均值, 除以方差, 因此数据的均值为0, 方差为1
print preprocessing.scale(a)

结果

1 2	[ -10. 2.3 13.7 56. 108. ] [-1.01951435 -0.73451375 -0.47036685 0.50975718 1.71463777]

Demo 2

import numpy as np
import matplotlib.pyplot as plt
from sklearn import preprocessing
from sklearn.cross_validation import train_test_split
from sklearn.datasets.samples_generator import make_classification
from sklearn.svm import SVC

# 生成数据集
X, y = make_classification(n_samples = 200, n_features = 2, n_redundant = 0, n_informative = 2, 
                           random_state = 22, n_clusters_per_class = 1, scale = 100)
# 查看数据分布
plt.scatter(X[:, 0], X[:, 1], c = y)
plt.show()

# 数据归一化处理, 不进行处理时注释掉
X = preprocessing.scale(X)
# 将数据分为训练集和测试集
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.2)
# 构建分类器
svm = SVC()
# 训练分类器
svm.fit(X_train, y_train)
# 测试
print svm.score(X_test, y_test)

结果

# 进行归一化的准确率
0.9
# 不进行归一化的准确率
0.65