格式:file1=pd.read_csv(file1_path)
pd.read_csv(file1_path,encoding='gbk')
pd.read_csv(file1_path,encoding='gbk',skiprows=[2,3])
pd.read_csv(file1_path,encoding='gbk',skiprows=lambda x:x%2==1)
pd.read_csv(file1_path,encoding='gbk',keep_default_na=False)
new=pd.DataFrame()
new.new[[0,1,2]]
new.new[0:2]
查詢結果同上
new.loc[new['激活數']>1000]
loc和iloc的區別:
loc:純標簽篩選
iloc:純數字篩選
#篩選出new的某兩列
new=new.loc[:,['phone','收件人姓名']]
#篩選new的第0,1列
new.iloc[:,[0,1]]
使用‘==’篩選-篩查“崔旭”的人(只能篩查指定明確的)
#new=file1.loc[(file1['收件人姓名']=='崔旭')|(file1['收件人姓名']=='崔霞')]
#print(new)
#使用loc函數篩選-str.contains函數-篩查名字中包含'亮'和'海'的人
#new=file1.loc[file1['收件人姓名'].str.contains('亮|海')]
#print(new)
#使用loc函數篩選-str.contains函數-篩查'崔'姓的人
#new=file1.loc[file1['收件人姓名'].str.startswitch('崔')]
#print(new)
df = df[(df['DEPOSIT_PAY_TIME_x'] .notnull() ) & (df['DEPOSIT_PAY_TIME_x'] != "" )]
print("during_time(number)=0的個數:",newdata[newdata['during_time(number)'] ==0].count()['during_time(number)'])
print("during_time(number)=1,2,3的個數:",newdata[(newdata['during_time(number)'] >0) & (newdata['during_time(number)'] <4)].count()['during_time(number)'])
print(newdata[newdata['during_time(number)'] ==0])
newdata[newdata['Team']. isin (['England','Italy','Russia'])][['Team','Shooting Accuracy']]
df.年齡.value_counts()
1.修改指定位置數據的值(修改第0行,’創建訂單數‘列的值為3836)
new.loc[0,'創建訂單數']=3836
2.替換‘小明’-‘xiaoming’
df.replace({'name':{'小明':'xiaoming'}})
3.批量替換某壹列的值(把‘性別’列裏的男-male,女-felmale)
方法壹:df['性別']=df['性別'].map({'男':'male','女':'female'})
方法二:df['性別'].replace('female','女',inplace=True)
或df['性別']=df['性別'].replace('female','女') 這就是inplace的作用
+df['性別'].replace('male','男',inplace=True)
4.替換列索引
df.columns=['sex','name','height','age']
或者:df.rename(columns={'性別':'sex','姓名':'name','身高':'height','年齡':'age'})
5.刪除某壹列
del df['player']
6. 刪除某壹列(方法二),刪除某壹行(默認axis=0刪除行,為1則刪除列)
刪除某壹列(方法二)
df.drop('性別',axis=1)
刪除某壹行
df.drop(1,axis=0)
file1=pd.read_csv(file1_path)
file2=pd.read_csv(file2_path)
new1=pd.DataFrame()
new1['phone']=file1['phone']
new1['contact_time']=file1['contact_time']
new2=pd.DataFrame()
new2['phone']=file2['phone']
new2['submission_audit_time']=file2['提交審核時間']
newdata=pd.merge(new1,new2,on='phone',how='left')
df=pd.concat([df1,df2],axis=0)
4.2.2 橫向表連接
df=pd.concat([df1,df2],axis=1)
df1['地區'].str.split('·',3,expand=True)
df1:
df1[['城市', '城區','地址']] = df1['地區'].str.split('·', 3, expand = True)
5.1 缺失值刪除
data.dropna(axis=0,subset = ["Age", "Sex"]) ? # 丟棄‘Age’和‘Sex’這兩列中有缺失值的行
data.dropna(how = 'all') # 傳入這個參數後將只丟棄全為缺失值的那些行
data.dropna(axis = 1) ? # 丟棄有缺失值的列(壹般不會這麽做,這樣會刪掉壹個特征)
data.dropna(axis=1,how="all") ? # 丟棄全為缺失值的那些列
5.2 缺失值填充:pandas.DataFrame.fillna()函數
DataFrame.fillna(value=None, method=None, axis=None, inplace=False, limit=None, downcast=None, **kwargs)
功能:使用指定方法填充NA/NaN值
其中inplace=True就是直接在原有基礎上填滿
5.3 缺失值查詢:
缺失值數量查詢:df.isnull().sum()
缺失值行查詢:df[df.isnull().T.any()]
newdata['during_time']=pd.to_datetime(newdata['submission_audit_time'])-pd.to_datetime(newdata['contact_time'])
newdata['during_time(number)']=(pd.to_datetime(newdata['submission_audit_time'])-pd.to_datetime(newdata['contact_time'])).apply(lambda x: x.days)
new=pd.DataFrame()
new=newdata[newdata['during_time(number)'] ==0]
new.to_csv(save_path,encoding='utf-8-sig')
將數據按行拆分並存儲到不同的csv文件中:
path='C:/Users/EDZ/Desktop/工作/2021.08.19/'
for i in range(0,30):
df.loc[[i]].to_csv(path+str(i)+'.csv',encoding='gbk')
df = df[['購藥日期', '星期','社保卡號','商品編碼', '商品名稱', '銷售數量', '應收金額', '實收金額' ]]