假设我们有两个数据帧:
df1 = pd.DataFrame({
0: 'ETERNITON',
1: 'CIELOON',
2: 'M.DIASBRANCOON',
3: 'IRBBRASIL REON',
4: '01/00 ATACADÃO S.A ON',
5: 'AMBEV S/A ON',
6: '01/00 RUMO S.A. ON',
7: 'COGNA ONON',
8: 'CURY S/A'}.items(), columns=['index', 'name']).set_index('index')
df2 = pd.DataFrame({'name': {0: 'ALLIARON', 1: 'M.DIASBRANCOON', 2: 'AMBEVS/AON', 3: 'CIELOON',
4: 'AESBRASILON', 5: 'BRASILAGROON', 6: 'IRBBRASILREON', 7: 'ATACADÃOS.AON', 8: 'ALPARGATASON',
9: 'RUMOS.A.ON', 10: 'COGNAONON'},
'yf_ticker': {0: 'AALR3.SA', 1: 'MDIA3.SA', 2: 'ABEV3.SA', 3: 'CIEL3.SA', 4: 'AESB3.SA',
5: 'AGRO3.SA', 6: 'IRBR3.SA', 7: 'CRFB3.SA', 8: 'ALPA3.SA', 9: 'RAIL3.SA', 10: 'COGN3.SA'}})
我想使用df2中的列"yf_ticker"在df1中创建一个新列("ticker"(。如果df2['yf_ticker']
中的名称/字符串在df1['name']
中(即使它不完全匹配(,则将df2中的yf_ticker添加到df1['ticker']
中的该行。为了明确起见,预期的输出将类似于:
print(df1)
name ticker
ETERNITON Missing or N/A or Nan
CIELOON CIEL3.SA
M.DIASBRANCOON MDIA3.SA
IRBBRASIL REON IRBR3.SA
01/00 ATACADÃO S.A ON CRFB3.SA
AMBEV S/A ON ABEV3.SA
01/00 RUMO S.A. ON RAIL3.SA
COGNA ONON COGN3.SA
CURY S/A Missing or N/A or Nan
我尝试过的解决方案:
df1['name'] = df1['name'].str.replace(" ","")
for i in range(len(df1)):
for j in range(len(df2)):
if df2.iloc[j,0] in df1.iloc[i,0]:
df1.loc[i, 'ticker'] = df2.iloc[j,1]
尽管它有效,但在我看来,对于更大的数据集,这样的for循环是低效的。有没有更快(或"矢量化"(的方法可以做到这一点?
我建议对name
列进行模糊匹配,然后从匹配行中获得yf_ticker
。下面是一个python内置difflib
:的例子
import difflib
df1['yf_ticker'] = df1['name'].apply(lambda x: df2.loc[df2['name'] == y[0], 'yf_ticker'].iloc[0] if (y := (difflib.get_close_matches(x, df2.name))) else None)
输出:
索引 | 名称 | yf_ticker |
---|---|---|
0 | ETERNITON | IELOON|
2 | M.DIASBRANCOON | >MDIA3.SA|
3 | IRBBRASIL REON | >IRBR3.SA|
4 | 01/00 ATACADâO S.A ON | CRFB3.SA |
6 | 01/00 RUMO美国ON | >td style="text-align:left;">RAIL3.SA|
7 | COGNA ONONON | >COGN3.SA|
8 | CURY S/A | /table>