将日期时间列表转换为 int

Question

I have a pandas column that looks like this:

df3[['days_till_all_arrive']]

Int64Index([-7, 92], dtype='int64')
Int64Index([-20], dtype='int64')
Int64Index([83, 83], dtype='int64')
Int64Index([83, 83], dtype='int64')
Int64Index([92], dtype='int64')
...
Int64Index([91], dtype='int64')
Int64Index([91], dtype='int64')
Int64Index([91], dtype='int64')
Int64Index([92], dtype='int64')
Int64Index([92], dtype='int64')

It gets created with a function

def days_till_arrival(df):
    res = []
    for item in df['ETA']:
        item = pd.to_datetime(item)
        res.append((item - datetime.now()).astype('timedelta64[D]'))
    
    return res

ETA is a list of date values. I use it like so:

df3['days_till_all_arrive'] = days_till_arrival(df3)

I want the values to be normal lists like so:

df3[['days_till_all_arrive']]

[-7, 92]
[-20]
[83, 83]
[83, 83]
[92]
...
[91]
[91]
[91]
[92]
[92]

I use .astype('timedelta64[D]') on a single value and it does convert it to the value I want of days, but when I use it in that function for multiple values I get more than [-7, 92], instead I get Int64Index([-7, 92], dtype='int64') .

How can I achieve that?

Answer 1

将list与Timedelta.days一起Timedelta.days ：

res.append(list((item - datetime.now()).days))