使用pandas模块实现数据的标准化操作

import pandas as pd
import numpy as np
def three_sigma(se):
    """
    自实现3σ原则，进行数据过滤
    :param se:传进来的series结构数据
    :return:去除异常值之后的series数据
    """
    bool_id=((se.mean()-3*se.std())<se) & (se<(se.mean()+3*se.std()))
    print(bool_id)
    return se[bool_id]
 
#加载数据
detail=pd.read_excel('./meal_order_detail.xlsx')
#进行异常值处理
res=three_sigma(detail['amounts'])
print(detail.shape)
print(res.shape)

(x-min)/(max-min)

import pandas as pd
import numpy as np
 
def minmax_sca(data):
    """
    离差标准化
    param data:传入的数据
    return:标准化之后的数据
    """
    new_data=(data-data.min())/(data.max()-data.min())
    return new_data
 
#加载数据
detail=pd.read_excel('./meal_order_detail.xlsx')
res=minmax_sca(detail[['amounts','counts']])
print(res)
data=res
bool_id=data.loc[:,'count']==1
print(data.loc[bool_id],'counts')

import pandas as pd
import numpy as np
def stand_sca(data):
    """
    标准差标准化
    :param data:传入的数据
    :return:标准化之后的数据
    """
    new_data=(data-data.mean())/data.std()
    return new_data
 
#加载数据
detail=pd.read_excel('./meal_order_detail.xlsx')
res=stand_sca(detail[['amounts','counts']])
print(res)
print('res的均值：',res.mean())
print('res的标准差：',res.std())

x/(10^k)
k=math.ceil(log10(max(|x|)))

import pandas as pd
import numpy as np
def deci_sca(data):
    """
    自实现小数定标标准化
    :param data: 传入的数据
    :return: 标准化之后的数据
    """
    new_data=data/(10**(np.ceil(np.log10(data.abs().max()))))
    return new_data
#加载数据
detail = pd.read_excel('./meal_order_detail.xlsx')
res = deci_sca(detail[['amounts', 'counts']])
print(res)

from sklearn import preprocessing
# 获取数据
cols = ['OverallQual','GrLivArea', 'GarageCars','TotalBsmtSF', 'FullBath', 'TotRmsAbvGrd', 'YearBuilt'] ##选取列
x = data_train[cols].values
y = data_train['SalePrice'].values  
x_scaled = preprocessing.StandardScaler().fit_transform(x) ##进行归一化
y_scaled = preprocessing.StandardScaler().fit_transform(y.reshape(-1,1))##先将y转换成一列，再进行归一
 

# 先将数据标准化
from sklearn.preprocessing import StandardScaler
ss = StandardScaler() ##
#用测试集训练并标准化
ss.fit(missing_age_X_train)##首先fit
missing_age_X_train = ss.transform(missing_age_X_train) #进行transform
missing_age_X_test = ss.transform(missing_age_X_test)

from sklearn import preprocessing
f_names = ['CentralAir', 'Neighborhood'] ##需要处理的数据标签 
for x in f_names:
    label = preprocessing.LabelEncoder()
    data_train[x] = label.fit_transform(data_train[x]) ##数据标准化

title_mapping = {"Mr": 1, "Miss": 2, "Mrs": 3, "Master": 4, "Rare": 5}#将标签对应数值
train_df['Title'] = train_df['Title'].map(title_mapping)#处理数据
train_df['Title'] = train_df['Title'].fillna(0)##将其余标签填充为0值

train_test.loc[train_test["Age"].isnull() ,"age_nan"] = 1 ##将标签转换成1
train_test.loc[train_test["Age"].notnull() ,"age_nan"] = 0##将此标签成为0
train_test = pd.get_dummies(train_test,columns=['age_nan']) ##columns决定哪几行分列处理，prefix参数是每列前缀