CODEWITHSUNDEEP
pythonpandasdatetimetimedelta
kms
kms

Reputation: 2024

Unsupported type for timedelta seconds component: numpy.int64

I am using the timedelta function from datetime package to convert seconds into hours and minutes format.

Here's an example: 

secs = timedelta(seconds = 30)
print(secs)

00:00:30

When is run this on a pandas column that I need to convert, I run into the unsupported type error. It looks like this happens when it the function encounters a '0' and is not sure how to handle it, I think. 

EmpComm = pd.DataFrame({
         'Duration (secs)':[3488, 2994,0, 0],
         'Duration (hh:mm)':['0:58:08','0:49:54',np.nan,np.nan],

})
print (EmpComm)
   Duration (secs) Duration (hh:mm)
0             3488          0:58:08
1             2994          0:49:54
2                0              NaN
3                0              NaN

Data sample

I am wondering why it isn't able to convert seconds = 0 to 00:00:00. Here's my code:

for i, row in EmpComm.iterrows():

    val = timedelta(seconds = EmpComm['Duration (secs)'].iloc[i])
    EmpComm['Duration (hh:mm)'][i] = val

Upvotes: 2

Views: 6983

Answers (1)

jezrael
jezrael

Reputation: 862661

Dont use loops with iterrows, because exist vectorized solution.

You need to_timedelta with parameter unit:

EmpComm = pd.DataFrame({
         'Duration (secs)':[3488, 2994,0, 0],
         'Duration (hh:mm)':['0:58:08','0:49:54',np.nan,np.nan],

})

EmpComm['Duration (hh:mm)'] = pd.to_timedelta(EmpComm['Duration (secs)'], unit='s')
print (EmpComm)
   Duration (secs) Duration (hh:mm)
0             3488         00:58:08
1             2994         00:49:54
2                0         00:00:00
3                0         00:00:00

Upvotes: 4

Related Questions