来自两个文本列的熊猫 to_datetime

Question

I have the following time-series of the format below.

What's the easiest way to convert/combine columns 'date' and 'time' into a pandas datetime format?

I know it should be pandas.to_datetime(date...). But I can't figure out the format to combine them.

  ccy      date  time    open    high     low   close
0  EURUSD  20190211   100  1.1318  1.1319  1.1317  1.1319
1  EURUSD  20190211   200  1.1320  1.1322  1.1319  1.1319
2  EURUSD  20190211   300  1.1318  1.1319  1.1318  1.1319
3  EURUSD  20190211   400  1.1319  1.1319  1.1318  1.1318
4  EURUSD  20190211   500  1.1318  1.1318  1.1318  1.1318

Answer 1

This is how you would do the conversion, but you need to make sure the time column makes sense - your table is ambiguous right now when displaying 3 digits for time. Also, make sure your date and time columns are strings, else convert them to strings because joining them sensibly will require it.

import pandas as pd
df=pd.DataFrame({'date':['20190211','20190211'],'time':['0100','0200']})
pd.to_datetime(df['date']  + df['time'], format='%Y%d%m%H%M')

Answer 2

It seems like date and time are currently stored as integers. Therefore you might need to cast them as strings in order to perform string slicing.

def convert_to_datetime(date, time):
    datestr = str(date)
    timestr = str(time)
    result = datestr[0:4] + "/" + datestr[4:6] + "/" + datestr[6:]
    result = result + " " + timestr[0:-2] + ":"+ timestr[-2:]
    return pd.to_datetime(result)

df["datetime"] = df.apply(lambda x: convert_to_datetime(x["date"], x["time"], axis=1)

Answer 3

Two ways to do it:

First method: (without pandas)

# convert date
data['ddate']= [str(dt)[:4]+'-'+str(dt)[4:6]+'-'+str(dt)[-2:] for dt in 
# convert time 
data['dtime']= [str(dt)[:-2]+':'+str(dt)[-2:] for dt in data['dtime']]
# join 
data['datetime'] = data['ddate'] + ' ' + data['dtime'] # space or whatever you want
# delete unneeded columns 
del data['ddate'], data['dtime']

Second method: (using pandas)

data['datetime']= pd.to_datetime([str(dt)+str(tm) for dt,tm in data[['date','time']].values],format='%Y%d%m%H%M')