ie)
count
2015-01 2
2015-02 1
2015-03 4
for the group i tried pd.groupby(b,by=[b.index.month,b.index.year])
but there was object has no attribute 'month' error
-
1Are those actual date objects or strings?piRSquared– piRSquared2017-04-18 18:49:58 +00:00Commented Apr 18, 2017 at 18:49
-
date is coloumn in dataframe objectsd.doo– d.doo2017-04-18 18:56:56 +00:00Commented Apr 18, 2017 at 18:56
Add a comment
|
1 Answer
Apply sorted with the key parameter set to pd.to_datetime
df.assign(date=df.date.apply(sorted, key=pd.to_datetime))
id date
0 a [2015-02-01, 2015-03-01]
1 b [2015-03-01]
2 s [2015-01-01, 2015-03-01]
3 f [2015-01-01, 2015-01-01, 2015-03-01]
Then use pd.value_counts
pd.value_counts(pd.to_datetime(df.date.sum()).strftime('%Y-%m'))
2015-03 4
2015-01 3
2015-02 1
dtype: int64
debugging
You should be able to copy and paste this code... please verify that it runs as expected.
import pandas as pd
df = pd.DataFrame(dict(
id=list('absf'),
date=[
['2015-03-01', '2015-02-01'],
['2015-03-01'],
['2015-01-01', '2015-03-01'],
['2015-01-01', '2015-01-01', '2015-03-01']
]
))[['id', 'date']]
print(df.assign(date=df.date.apply(sorted, key=pd.to_datetime)))
print()
print(pd.value_counts(pd.to_datetime(df.date.sum()).strftime('%Y-%m')))
You should expect to see
id date
0 a [2015-02-01, 2015-03-01]
1 b [2015-03-01]
2 s [2015-01-01, 2015-03-01]
3 f [2015-01-01, 2015-01-01, 2015-03-01]
2015-03 4
2015-01 3
2015-02 1
dtype: int64
5 Comments
piRSquared
I don't understand what the problem is? You showed a dataframe
df. Did you run this code on df? You also did not give enough information to tell what is going on.piRSquared
@d.doo see the debugging section of my updated answer.
d.doo
df.assign(dt=df.dt.apply(sorted, key=pd.to_datetime)) there was TypeError: Unrecognized value type: <class 'str'> and ValueError: Given date string not likely a datetime.
piRSquared
Can you edit your question with the full stack trace?
Simon
Debugging code works ok for me pd.__version__ 0.19.2 - d.doo could you just confirm if the debugging code works correctly please. The problem may be with your source data. pd.to_datetime also has errors='coerce' and errors='ignore' that you could try.