Question

所以我有一个看起来像这样的df：

Created UserID  Service
1/1/2016    a   CWS
1/2/2016    a   Other
3/5/2016    a   Drive
2/7/2017    b   Enhancement
... ... ...

我想根据＆＃34;服务＆＃34;中的值来过滤它。 CWS和Drive的列。我是这样做的：

df=df[(df.Service=="CWS") or (df.Service=="Drive")]

它无法正常工作。有什么想法吗？

Answer 1

需要与|（or）进行比较：

df=df[(df.Service=="CWS") | (df.Service=="Drive")]

更好的是使用isin：

df=df[(df.Service.isin(["CWS", "Drive")]])

或使用query：

df = df.query('Service=="CWS" | Service=="Drive"')

或query with list：

df = df.query('Service== ["Other", "Drive"]')

print (df)
    Created UserID Service
1  1/2/2016      a   Other
2  3/5/2016      a   Drive

Answer 2

您也可以使用pandas.Series.str.match

df[df.Service.str.match('CWS|Drive')]

    Created UserID Service
0  1/1/2016      a     CWS
2  3/5/2016      a   Drive

其他口味
好玩!!

<强> numpy -fi

s = df.Service.values
c1 = s == 'CWS'
c2 = s == 'Drive'
df[c1 | c2]

添加numexpr

import numexpr as ne

s = df.Service.values
c1 = s == 'CWS'
c2 = s == 'Drive'
df[ne.evaluate('c1 | c2')]

<强>时序
isin是赢家！ str.match是输家： - （

np.random.seed([3,1415])
df = pd.DataFrame(dict(
        Service=np.random.choice(['CWS', 'Drive', 'Other', 'Enhancement'], 100000)))

%timeit df[(df.Service == "CWS") | (df.Service == "Drive")]
%timeit df[df.Service.isin(["CWS", "Drive"])]
%timeit df.query('Service == "CWS" | Service == "Drive"')
%timeit df.query('Service == ["Other", "Drive"]')
%timeit df.query('Service in ["Other", "Drive"]') 
%timeit df[df.Service.str.match('CWS|Drive')]

100 loops, best of 3: 16.7 ms per loop
100 loops, best of 3: 4.46 ms per loop
100 loops, best of 3: 7.74 ms per loop
100 loops, best of 3: 5.77 ms per loop
100 loops, best of 3: 5.69 ms per loop
10 loops, best of 3: 67.3 ms per loop

%%timeit
s = df.Service.values
c1 = s == 'CWS'
c2 = s == 'Drive'
df[c1 | c2]

100 loops, best of 3: 5.47 ms per loop

%%timeit 
import numexpr as ne

s = df.Service.values
c1 = s == 'CWS'
c2 = s == 'Drive'
df[ne.evaluate('c1 | c2')]

100 loops, best of 3: 5.65 ms per loop

根据列中的两个可能值过滤Panda DF

2 个答案: