如何在 pandas 的数据框中使用正则表达式提取部分文本

Question

I have a data frame and one of the columns are like this:

df = index  dosage_duration
     0        5  years20mg 1X D
     1         2  days10mg 1X D
     2         2  days10mg 1X D
     3                 7  weeks
     4                2  months
     5                  3  days
     6             1  years5 MG
     7                 2  years

What I am trying to do is to extract the first part of the time and convert it to days. So the result would look like this:

df = index  dosage_duration       new_dosage
     0        5  years20mg 1X D    5*365
     1         2  days10mg 1X D    2
     2         2  days10mg 1X D    2
     3                 7  weeks    7*7
     4                2  months    2*30
     5                  3  days    3
     6             1  years5 MG    1*365
     7                 2  years    2*365

As you see here, 5 years being converted to 5*365 to be as days.

I am able to get the first part lets to say 5 in the first row , 2 in the second row... but Im not sure how can I get the years days or month so I can change all the values to days scale.

Apparently, I need to be able to find the first number after the space but I don't know how can I do this part.

Answer 1

Let's try:

df = pd.DataFrame({'dosage_duration':['5 years20mg 1x D'
                                     ,'2 days10mg 1x D'
                                     ,'4 months20mg 1x D'
                                     ,'7 weeks'
                                     ,'2 months'
                                     ,'3 days'
                                     ,'1 days'
                                     ,'1 years5 MG'
                                     ,'2 years'
                                     ,'6 months'
                                     ,'1 years10 1x D'
                                     ,'10 months15']})

nmap={'years':365, 'months':30, 'weeks':7, 'days': 1}
strnmap = '|'.join(nmap.keys())

df_m = df.dosage_duration.str.extract(f'(?P<unit>\d+)\s?(?P<span>[{strnmap}]+)')
df['new_duration']= df_m['unit'].astype(int).mul(df_m['span'].map(nmap))

print(df)

Output:

      dosage_duration  new_duration
0    5 years20mg 1x D          1825
1     2 days10mg 1x D             2
2   4 months20mg 1x D           120
3             7 weeks            49
4            2 months            60
5              3 days             3
6              1 days             1
7         1 years5 MG           365
8             2 years           730
9            6 months           180
10     1 years10 1x D           365
11        10 months15           300
​

Answer 2

• split by space.
• the first element is your number.
• The second element indicates what kind of time it is? day , week , month , year . Just the first letter is enough to identify what to multiply.

import pandas as pd

df  = pd.DataFrame({'dosage_duration':['5 years27abc','10 days92pqr', '5.5 weeks782364hgsdf', '3 months21647hadjh']})

mul = {
    'd':1,
    'w':7,
    'm':30,
    'y':365
}

df['new_dosage'] = df['dosage_duration'].apply(lambda x:x.split()).apply(lambda x:float(x[0])*mul[x[1][0]])
df

Output:

    dosage_duration     new_dosage
0   5 years27abc        1825
1   10 days92pqr        10
2   5.5 weeks782364hgsdf    35
3   3 months21647hadjh  90

Update:

• if you want them as string of expression.

import pandas as pd

df  = pd.DataFrame({'t':['5 years27abc','10 days92pqr', '5 weeks782364hgsdf', '3 months21647hadjh']})

mul = {
    'd':'1',
    'w':'7',
    'm':'30',
    'y':'365'
}

df['total_time'] = df['t'].apply(lambda x:x.split()).apply(lambda x:x[0] + '*' + mul[x[1][0]])
df

Output:

          t             total_time
0   5 years27abc        5*365
1   10 days92pqr        10*1
2   5 weeks782364hgsdf  5*7
3   3 months21647hadjh  3*30