Question

我试图找到日期时间格式的pandas数据框中两列之间的时间差异。

以下是我的数据框中的一些数据以及我一直使用的代码。我已经三次检查这两个列dtypes是datetime64。

我的数据：

date_updated                  date_scored 
2016-03-30 08:00:00.000       2016-03-30 08:00:57.416  
2016-04-07 23:50:00.000       2016-04-07 23:50:12.036

我的代码：

data['date_updated'] = pd.to_datetime(data['date_updated'], 
format='%Y-%m-%d %H:%M:%S')
data['date_scored'] = pd.to_datetime(data['date_scored'], 
format='%Y-%m-%d %H:%M:%S')
data['Diff'] =  data['date_updated'] - data['date_scored']

我收到的错误讯息：

TypeError: data type "datetime" not understood

任何帮助将不胜感激，谢谢！

我的解决方案：

for i in raw_data[:10]:
scored = i.date_scored
scored_date =  pd.to_datetime(scored, format='%Y-%m-%d %H:%M:%S')
if type(scored_date) == "NoneType":
    pass
elif scored_date.year >= 2016:
    extracted = i.date_extracted
    extracted =  pd.to_datetime(extracted, format='%Y-%m-%d %H:%M:%S')
    bank = i.bank.name
    diff = scored - extracted
    datum = [str(bank), str(extracted), str(scored), str(diff)]
    data.append(datum)
else:
    pass

Answer 1

我使用上面的语法遇到了同样的错误（虽然在另一台机器上工作）：

data['Diff'] =  data['date_updated'] - data['date_scored']

它在我的新机器上工作：

data['Diff'] =  data['date_updated'].subtract(data['date_scored'])

Answer 2

它就像一个魅力。您甚至可以简化代码，因为to_datetime足够智能，可以为您猜测格式。

import io
import pandas as pd
# Paste the text by using of triple-quotes to span String literals on multiple lines
zz = """date_updated,date_scored
2016-03-30 08:00:00.000,       2016-03-30 08:00:57.416  
2016-04-07 23:50:00.000,       2016-04-07 23:50:12.036"""

data = pd.read_table(io.StringIO(zz), delim_whitespace=False, delimiter=',')

data['date_updated'] = pd.to_datetime(data['date_updated'])
data['date_scored'] = pd.to_datetime(data['date_scored'])
data['Diff'] =  data['date_updated'] - data['date_scored']

print(data)
#          date_updated             date_scored                     Diff
# 0 2016-03-30 08:00:00 2016-03-30 08:00:57.416 -1 days +23:59:02.584000
# 1 2016-04-07 23:50:00 2016-04-07 23:50:12.036 -1 days +23:59:47.964000

Answer 3

You need to update pandas. I've just ran into the same issue with an old code that used to run without issues. After updating pandas (0.18.1-np111py35_0) to a newer version (0.20.2-np113py35_0) the issue was resolved.

pandas数据帧中的日期时间不会相互减去

3 个答案: