[英]Sum of Timedeltas column in Pandas Python
但是,當我嘗試所有可用的解決方案時,它給了我錯誤。
碼:
print sum(data['Activity_Duration'],datetime.timedelta())
#import operator
#print reduce(operator.add, data['Activity_Duration'])
錯誤:
OverflowError
1 #print sum(data ['Activity_Duration'],datetime.timedelta())
2名進口經營者
----> 3打印reduce(operator.add,data ['Activity_Duration'])OverflowError:太大而無法轉換
我是否缺少某些東西,或者我們可以提出一個更具擴展性的解決方案?
信息:我的數據有436746行。
我正在使用8 gig機器,數據大小為650MB
我認為你需要sum
:
print (df['Activity_Duration'].sum())
樣品:
import pandas as pd
start = pd.to_datetime('2015-02-24')
end = pd.to_datetime('2016-04-25')
rng = pd.date_range(start, end, freq='6D')
start = pd.to_datetime('2015-02-26')
end = pd.to_datetime('2016-04-27')
rng1 = pd.date_range(start, end, freq='6D')
df = pd.DataFrame({'Date1': rng, 'Date2': rng1})
df['Activity_Duration'] = df.Date2 - df.Date1
print (df)
Date1 Date2 Activity_Duration
0 2015-02-24 2015-02-26 2 days
1 2015-03-02 2015-03-04 2 days
2 2015-03-08 2015-03-10 2 days
3 2015-03-14 2015-03-16 2 days
4 2015-03-20 2015-03-22 2 days
5 2015-03-26 2015-03-28 2 days
6 2015-04-01 2015-04-03 2 days
7 2015-04-07 2015-04-09 2 days
8 2015-04-13 2015-04-15 2 days
9 2015-04-19 2015-04-21 2 days
...
...
print (df['Activity_Duration'].sum())
144 days 00:00:00
如果需要在float
輸出:
import numpy as np
df['Activity_Duration'] = (df.Date2 - df.Date1) / np.timedelta64(1, 'D')
print (df)
Date1 Date2 Activity_Duration
0 2015-02-24 2015-02-26 2.0
1 2015-03-02 2015-03-04 2.0
2 2015-03-08 2015-03-10 2.0
3 2015-03-14 2015-03-16 2.0
4 2015-03-20 2015-03-22 2.0
...
...
...
print (df['Activity_Duration'].sum())
144.0
另一個解決方案是dt.days
輸出是int
:
print (df['Activity_Duration'].dt.days.sum())
144
聲明:本站的技術帖子網頁,遵循CC BY-SA 4.0協議,如果您需要轉載,請注明本站網址或者原文地址。任何問題請咨詢:yoyou2525@163.com.