python数据预处理

一、数据规范化

 

import pymysql

import pandas as pd

import numpy as np

conn = pymysql.connect(host="",user="root",passwd="root",db="csdn")

sql = "select price,comment from taob"

data=pd.read_sql(sql,conn)

#离差标准化

data2 = (data-data.min())/(data.max() - data.min())

# 标准差标准化

data3 = (data-data.mean())/data.std()

# 小数定标规范化

k = np.ceil(np.log10(data.abs().max()))

data4 = data / 10**k

 

 

二、离散化

 

连续数据离散化

# 等宽离散化

data5 = data[u"price"].copy()

data6 = data5.T

data7 = data6.values

k =3

c1 = pd.cut(data7,k,labels = ["便宜","适中","贵"])

# 非等宽离散化

k1 = [0,50,100,300,500,2000,data7.max()]

c2 = pd.cut(data7,k1)

# 等频率离散化

# 一维聚类离散