1. Min-max 规范化
# coding:utf-8
from sklearn import preprocessing
import numpy as np
# 初始化数据,每一行表示一个样本,每一列表示一个特征
x = np.array([[ 0., -3., 1.],
[ 3., 1., 2.],
[ 0., 1., -1.]])
# 将数据进行[0,1]规范化
min_max_scaler = preprocessing.MinMaxScaler()
minmax_x = min_max_scaler.fit_transform(x)
print(minmax_x)
2. Z-Score 规范化
from sklearn import preprocessing
import numpy as np
# 初始化数据
x = np.array([[ 0., -3., 1.],
[ 3., 1., 2.],
[ 0., 1., -1.]])
# 将数据进行Z-Score规范化
scaled_x = preprocessing.scale(x)
print(scaled_x)
3. 小数定标规范化
# coding:utf-8
from sklearn import preprocessing
import numpy as np
# 初始化数据
x = np.array([[ 0., -3., 1.],
[ 3., 1., 2.],
[ 0., 1., -1.]])
# 小数定标规范化
j = np.ceil(np.log10(np.max(abs(x))))
scaled_x = x/(10**j)
print(scaled_x)