多场景下的Pandas使用技巧分享

df_source = pd.read_csv('[file_path].csv', encoding='UTF-8')

df_source = pd.read_excel('[file_path].xlsx', encoding='utf-8')  # 原始dataframe

df_source = pd.read_excel('[file_path].xlsx', encoding='utf-8')  # 原始dataframe
df_extract_left = df_source['[col_name]']

df_source = pd.read_excel('[file_path].xlsx', encoding='utf-8')
col_n = ['col_name1', 'col_name2', 'col_name3', 'col_name4']
df_extract_right = pd.DataFrame(df_source ,columns=col_n)

df_merge = pd.merge(df_left, df_right, how='left', left_on='[左表的列名]', right_on='【右表的列名】')

left_dict = {
    'Fruits':['Apple','Banana','Cherry','Dates','Eggfruit'],
    'Color': ['Red', 'Gray', 'Red', 'Brown', 'Yellow']
}
df_left = pd.DataFrame(left_dict)
df_left

right_dict = {
    'Fruits':['A','B','C','D','E'],
    'Color': ['Red', 'Yellow', 'Red', 'Brown', 'Yellow']
}
df_right = pd.DataFrame(right_dict)
df_right

df = pd.merge(df_left, df_right, how='left', left_on='Color', right_on='Color')
df

# 去除列多余的字符 \r\n
df['要处理列的列名'] = df.apply(lambda x:str(x['要处理列的列名']).replace('\r','').replace('\n','') ,axis=1)
df

df2 = df.groupby('group by的列')['做字符串拼接的列'].apply(lambda x:x.str.cat(sep=","))
df2

df.groupby(by=['聚合列1','聚合列2'])['第三列'].sum().reset_index(drop=False)

df.drop([0, 1])

df.drop(labels=[‘id',‘class'], axis=1)

df[df['列名'] == '待筛选的值']

def concat_df(x):
    df_res = pd.pivot(x, index='col_a', columns='col_b', values="col_c")
    df_res.insert(0,'col_a',x.iloc[0,0]) #调试时发现结果缺失col_a列，故添加该列值
    return df_res
df_res =  df.groupby('col_a').apply(lambda x : concat_df(x))

df_res=df.drop(df[df['列名']=='条件'].index)

for name,df_group in df.groupby(by=['key']):
	# name为当前group的key值 str类型
	# df_group为当前的子group dataframe类型

# x为每个group的dataframe
name,df_group in df.groupby(by=['key']).apply(lamdba x : print(x))

df2 = pd.DataFrame([['Tom', 16], ['Nancy', 18], ['Jack', 15],['Tom', 18],['Jack', 13],['Jack', 115]],columns=['cola','colb'])
	
	cola	colb
	Tom	    16
	Nancy	18
	Jack	15
	Tom	    18
	Jack	13
	Jack	115

# 方法一:
for name,df_group in df2.groupby(by=['cola']):
    # 只演示第一个group的值
    print(name) # Jack
    print(df_group)
   cola  colb
   Jack    15
   Jack    13
   Jack   115

# 方法二
df2.groupby(by=['cola']).apply(lambda x: print(x))
  
   cola  colb
2  Jack    15
4  Jack    13
5  Jack   115
   
    cola  colb
1  Nancy    18
 
  cola  colb
0  Tom    16
3  Tom    18

df.rename(columns={'two':'twotwo'},inplace=True)

		df_res = pd.DataFrame()
		df['DATE1'] = df['DATE'].shift(1)
        df_time = (df['DATE'] - df['DATE1']).dt.total_seconds()
        df_time.dropna(axis=0,inplace=True)
        df_res = pd.concat([df_res,df_time])

df.dropna(axis=0,inplace=True) # axis = 0 为行 axis = 1为列 ，inplace = True表示替换原df

df2=df2[~df2['a'].isin(df1['a'].unique())]

count_min = min(df['列名'].value_counts())
# 返回一个series

len(set(df1['a']) & set(df1['b']))

df1.loc[~(df1['a']==df1['b'])]

df.loc[df['a']==True,'b'].sum()