Question

我有一个像这样的pandas数据框：

SUBJECT                 STUDENT         CITY        STATE

Math/Chemistry/Biology  Sam/Peter/Mary  Los Angeles CA
Geology/Physics         John            Boston      MA

这应该是这样的：

SUBJECT      STUDENT    CITY           STATE

Math         Sam        Los Angeles    CA
Chemistry    Peter      Los Angeles    CA
Biology      Mary       Los Angeles    CA
Geology      John       Boston         MA
Physics      John       Boston         MA

在提出这个问题之前，我提到了本页提到的解决方案： pandas: How do I split text in a column into multiple rows?

由于两列中存在斜杠分隔的字符串，因此我无法使用上述链接中的解决方案。

Answer 1

concat和join的另一种解决方案：

s1 = df.SUBJECT.str.split('/', expand=True).stack()
s2 = df.STUDENT.str.split('/', expand=True).stack()
print (s1)
0  0         Math
   1    Chemistry
   2      Biology
1  0      Geology
   1      Physics

print (s2)
0  0      Sam
   1    Peter
   2     Mary
1  0     John
dtype: object

df1 = pd.concat([s1,s2], axis=1, keys=('SUBJECT','STUDENT'))
        .ffill()
        .reset_index(level=1, drop=True)
print (df1)
     SUBJECT STUDENT
0       Math     Sam
0  Chemistry   Peter
0    Biology    Mary
1    Geology    John
1    Physics    John

df = df.drop(['SUBJECT','STUDENT'], axis=1)
       .join(df1)
       .reset_index(drop=True)[['SUBJECT', 'STUDENT', 'CITY','STATE']]
print (df)
     SUBJECT STUDENT         CITY STATE
0       Math     Sam  Los Angeles    CA
1  Chemistry   Peter  Los Angeles    CA
2    Biology    Mary  Los Angeles    CA
3    Geology    John       Boston    MA
4    Physics    John       Boston    MA

Answer 2

首先，按'/'

分割字段

df.SUBJECT = df.SUBJECT.str.split('/')
df.STUDENT = df.STUDENT.str.split('/')

然后我使用一个函数来爆炸行。但是，我不得不隔离那些只有一个学生或主题的行。

def explode(df, columns):
    idx = np.repeat(df.index, df[columns[0]].str.len())
    a = df.T.reindex_axis(columns).values
    concat = np.concatenate([np.concatenate(a[i]) for i in range(a.shape[0])])
    p = pd.DataFrame(concat.reshape(a.shape[0], -1).T, idx, columns)
    return pd.concat([df.drop(columns, axis=1), p], axis=1).reset_index(drop=True)


cond = df.STUDENT.str.len() == df.SUBJECT.str.len()

df_paired = df[cond]
df_unpard = df[~cond]

if not df_paired.empty:
    df_paired = explode(df_paired, ['STUDENT','SUBJECT'])

if not df_unpard.empty:
    df_unpard = explode(explode(df_unpard, ['STUDENT']), ['SUBJECT'])

最后

pd.concat([df_paired, df_unpard], ignore_index=True)[df.columns]

时序

<强> piRSquared

%%timeit

df = df_.copy()

df.SUBJECT = df.SUBJECT.str.split('/')
df.STUDENT = df.STUDENT.str.split('/')

def explode(df, columns):
    idx = np.repeat(df.index, df[columns[0]].str.len())
    a = df.T.reindex_axis(columns).values
    concat = np.concatenate([np.concatenate(a[i]) for i in range(a.shape[0])])
    p = pd.DataFrame(concat.reshape(a.shape[0], -1).T, idx, columns)
    return pd.concat([df.drop(columns, axis=1), p], axis=1).reset_index(drop=True)

cond = df.STUDENT.str.len() == df.SUBJECT.str.len()

df_paired = df[cond]
df_unpard = df[~cond]

if not df_paired.empty:
    df_paired = explode(df_paired, ['STUDENT','SUBJECT'])

if not df_unpard.empty:
    df_unpard = explode(explode(df_unpard, ['STUDENT']), ['SUBJECT'])

pd.concat([df_paired, df_unpard], ignore_index=True)[df.columns]

100 loops, best of 3: 7.76 ms per loop

<强> jezrael

%%timeit

df = df_.copy()

s1 = df.SUBJECT.str.split('/', expand=True).stack()
s2 = df.STUDENT.str.split('/', expand=True).stack()

df1 = pd.concat([s1,s2], axis=1, keys=('SUBJECT','STUDENT')) \
        .ffill() \
        .reset_index(level=1, drop=True)

df.drop(['SUBJECT','STUDENT'], axis=1) \
  .join(df1) \
  .reset_index(drop=True)[['SUBJECT', 'STUDENT', 'CITY','STATE']]

100 loops, best of 3: 5.13 ms per loop

Answer 3

试试这个：可以修改SUBJECT等于1，然后使用zip。

df3.SUBJECT = df3.SUBJECT.str.split('/')
df3.STUDENT = df3.STUDENT.str.split('/')

def splitter(gb):
    ll = []
    subs, stus = gb.SUBJECT.values[0], gb.STUDENT.values[0]

    if   len(stus) == len(subs): ll = zip(subs,stus)
    elif len(stus) == 1:         ll = zip(subs,stus*len(subs))
    return pd.DataFrame(ll, columns= (["SUBJECT","STUDENT"])) 

df = df3.groupby(['CITY','STATE'])['SUBJECT','STUDENT'].apply(splitter).reset_index().drop('level_2', axis =1)
print df[[ 'SUBJECT', 'STUDENT', 'CITY','STATE' ]]


   SUBJECT STUDENT        CITY STATE
0    Geology    John      Boston    MA
1    Physics    John      Boston    MA
2       Math     Sam  LosAngeles    CA
3  Chemistry   Peter  LosAngeles    CA
4    Biology    Mary  LosAngeles    CA

Python Pandas：将两列或多列中的字符串斜线拆分为多行

3 个答案:

时序