利用python进行常见的数据预处理，主要是通过sklearn的preprocessing模块以及自写的方法来简单介绍下：

加载包及导入数据

# -*- coding:utf-8 -*-
import math
import numpy as np
from sklearn import datasets
from sklearn import preprocessing

iris = datasets.load_iris()
iris_X = iris.data[:4]
iris_y = iris.target[:4]

z-score标准化方法，均值为0，方差为1

# 方法1.sklearn下方法多列处理
def scale_encoder(iris_X):
    X = preprocessing.scale(iris_X)
    print(X[:4],'\n',np.mean(X),np.std(X))
    return X
f1 = scale_encoder(iris_X)
# 方法2.单列处理
def zscore_encoder(colvalue):
    average = float(sum(colvalue)) / len(colvalue)
    X = [(x - average) / np.std(colvalue) for x in colvalue]
    print(X[:4],'\n',np.mean(X),np.std(X))
    return X
f2 = zscore_encoder(iris_X[:,0])

min-max标准化（Min-Max Normalization）

# 方法1：处理多列
def min_max_scaler(iris_X):
    min_max_scaler = preprocessing.MinMaxScaler()
    X2 = min_max_scaler.fit_transform(iris_X)
    return X2
min_max_scaler(iris_X)
# 方法2：处理单列
def min_max_encoder(colvalue):
    new_value = [(x - min(colvalue))/(max(colvalue) - min(colvalue)) for x in colvalue]
    return new_value
min_max_encoder(iris_X[:,0])

规范化（Normalization），其中L2 norm：平方和为1；L1 norm：绝对值和为1

def normalize_encoder(iris_X):
    X3 = preprocessing.normalize(iris_X, norm='l2')
    X4 = preprocessing.normalize(iris_X, norm='l1')
    return X3
normalize_encoder(iris_X)

均值归一化方法[-1,1]

def average_max_min(data):
    average = float(sum(data)) / len(data)
    new_value = [(x - average) / (max(data) - min(data)) for x in data]
    return new_value
average_max_min(iris_X[:,0])

特征二值化（Binarization）,以某个阈值作为分割点进行切分数据

def binary_encoder(iris_X):
    binarizer = preprocessing.Binarizer(threshold=2)	#本例以阈值2为例
    X5 = binarizer.transform(iris_X)
    return X5
binary_encoder(iris_X)

标签二值化（Label binarization）

def label_encoder(iris_y):
    lb = preprocessing.LabelBinarizer()
    y2 = lb.fit_transform(iris_y)
    return y2
label_encoder(iris_y)

数值或字符标签化

def label_encoder(data):
    le = preprocessing.LabelEncoder()
    le.fit(data)
    t = le.transform(data)
    return t
label_encoder(iris_X[:,0])

独热编码

def onehot_encoder(data):
    enc = preprocessing.OneHotEncoder(sparse = False,categories='auto')
    t = enc.fit_transform(data)
    return t
onehot_encoder(iris_X)

对数变换

def log_encoder(col,logn):
    if logn == 2:
        new_value = [math.log2(x) for x in col]
    elif logn == 10:
        new_value = [math.log10(x) for x in col]
    else:
        print('暂不支持')
    return new_value
log_encoder(iris_X[:,0],logn=10)