Python用pandasによるデータ洗浄処理

1.データ読み出し

import pandas as pd
import numpy as np
import pymongo

data = pd.DataFrame(pd.read_excel('000.xlsx', index=False))

client = pymongo.MongoClient("mongodb://XX:[email protected]:2018",connect=False)
db = client["test"]
table = db["python"]
df = pd.DataFrame(list(table.find()))

 
EXcel,csv,mongoデータなどからデータを読み取ることができます
2.遍歴

for i in range(data.index.max()):  
    if any([  
        'missing' in data.loc[i,:].values,  
        data.loc[i,'hour'] not in range(25),  
        ]):  
  
        print('         %s    '%i)  
        data.drop([i],inplace=True)
for i in range(0,len(df)):
    info = df.loc[i].to_dict()

3.暇を取る(NA)
 
3.1直接除去

from numpy import nan as NA
data=Series([1,NA,3.5,NA,7])
print(data.dropna())
#  2 NA   
print(data.dropna(thresh=2))

3.2中位数または平均数で埋める

df = df.fillna(df.median())
print(df.fillna(df.mean()))

4.フィールドの処理
 

def get_salary(salary):
    s = 0
    if "-" in salary:
        for part in salary.split("-"):
            if " " in part:
                q = float(part[:-1]) * 10000
            else:
                q = float(part[:-1]) * 1000
            s += q
        return int(s/2.0)
    else:
        return np.nan    
df["salary"] = df["salary"].apply(get_salary)
df.head()
df["company"]=df["company"].apply(lambda x :x.split("/")[0].strip('"'))

5.重複除外
 

df["company"].drop_duplicates()

6.部分だけ残しておく

df.loc[:,["address","company"]]
df_c = df_c.iloc[:,[4,5]]
del data["name_grade"]
del data["info_grade"]

7.ソート

df.sort_values(by='col1', ascending=False)

8. isin

mask = df['A'].isin([1]) #      list

9. merge

df1 = pd.DataFrame({'name':['kate', 'herz', 'catherine', 'sally'], 'age':[25, 28, 39, 35]})
df2 = pd.DataFrame({'name_t':['kate', 'herz', 'catherine', 'sally'], 'score':[70, 60, 90, 100]})
print(pd.merge(df1, df2, left_on="name", right_on="name_t").drop('name_t', axis=1))

10.csvとして保存するか、mongoに保存する

df["company"].drop_duplicates().to_csv("company.csv",encoding="utf-8")
db[MONGO_TABLE].insert(row.to_dict()) 

 
http://www.codeblogbt.com/archives/102061