TypeError: unsupported operand type(s) for -: 'str' and 'pandas.

I've created a class which takes in minute data and returns the daily ohlc for that day. A simple version looks like so:

import pandas as pd
from datetime import time
from IPython.display import display
from pandas.tseries.holiday import USFederalHolidayCalendar
from pandas.tseries.offsets import CustomBusinessDay

US_BUSINESS_DAY = CustomBusinessDay(calendar=USFederalHolidayCalendar())

class SessionData:
    def __init__(self, data, date):
        self.data =  pd.read_csv(data)
        self.date = date

        df = self.data
        display(df)
        # get the minute data and return only the specified date (2022-04-18)
        df_current_day = df[(df['date'] >= date) & (df['date'] <= date)]
        df_current_day['time'] = pd.to_datetime(df['time']).dt.time

        self.previous_day = date - 2 * US_BUSINESS_DAY # 2018-7-2
        # get the minute data aof previous trading day (2022-04-14)
        df_previous_day = df[(df['date'] >= self.previous_day) & (df['date'] <= self.previous_day)]

Here's what my data originally looks like:

    v   vw  o   c   h   l   t   n   date    time
0   605.0   4.2036  4.2000  4.20    4.2000  4.20    2022-04-07 13:30:00 3   2022-04-07  13:30:00
1   809.0   4.2013  4.2026  4.20    4.2026  4.20    2022-04-07 13:41:00 12  2022-04-07  13:41:00
2   115.0   4.1739  4.1700  4.17    4.1700  4.17    2022-04-07 13:43:00 3   2022-04-07  13:43:00
3   170.0   4.1495  4.1500  4.15    4.1500  4.15    2022-04-07 13:53:00 6   2022-04-07  13:53:00
4   100.0   4.1600  4.1600  4.16    4.1600  4.16    2022-04-07 13:57:00 1   2022-04-07  13:57:00
... ... ... ... ... ... ... ... ... ... ...
1397    6260.0  6.5252  6.5300  6.53    6.5600  6.51    2022-04-18 23:55:00 32  2022-04-18  23:55:00
1398    8610.0  6.5399  6.5300  6.55    6.5500  6.52    2022-04-18 23:56:00 28  2022-04-18  23:56:00
1399    9035.0  6.5493  6.5500  6.55    6.5600  6.54    2022-04-18 23:57:00 24  2022-04-18  23:57:00
1400    30328.0 6.5188  6.5600  6.50    6.5600  6.50    2022-04-18 23:58:00 66  2022-04-18  23:58:00
1401    25403.0 6.5152  6.5000  6.52    6.5500  6.49    2022-04-18 23:59:00 62  2022-04-18  23:59:00
1402 rows × 10 columns

When I get the current date it's fine:

v   vw  o   c   h   l   t   n   date    time
687 852.0   4.1498  3.98    4.41    4.4100  3.98    2022-04-18 12:00:00 13  2022-04-18  12:00:00
688 2901.0  4.4839  4.13    4.75    4.7500  4.13    2022-04-18 12:01:00 24  2022-04-18  12:01:00
689 44063.0 4.9450  4.88    4.66    5.2599  4.60    2022-04-18 12:02:00 236 2022-04-18  12:02:00
690 46314.0 4.6890  4.70    4.62    4.8000  4.55    2022-04-18 12:03:00 225 2022-04-18  12:03:00
691 142991.0    4.8611  4.66    5.03    5.0900  4.61    2022-04-18 12:04:00 581 2022-04-18  12:04:00
... ... ... ... ... ... ... ... ... ... ...
1397    6260.0  6.5252  6.53    6.53    6.5600  6.51    2022-04-18 23:55:00 32  2022-04-18  23:55:00
1398    8610.0  6.5399  6.53    6.55    6.5500  6.52    2022-04-18 23:56:00 28  2022-04-18  23:56:00
1399    9035.0  6.5493  6.55    6.55    6.5600  6.54    2022-04-18 23:57:00 24  2022-04-18  23:57:00
1400    30328.0 6.5188  6.56    6.50    6.5600  6.50    2022-04-18 23:58:00 66  2022-04-18  23:58:00
1401    25403.0 6.5152  6.50    6.52    6.5500  6.49    2022-04-18 23:59:00 62  2022-04-18  23:59:00

But when I go to create a new dataframe based off the previous business day I get the following error:

---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
<ipython-input-346-dd4f43872f9b> in <module>
    182 #     # print(pm)
    183 
--> 184 pmData = SessionData('data\CASA.csv', '2022-04-18')
    185 
    186 # pmData.get_trading_session_times()

<ipython-input-346-dd4f43872f9b> in __init__(self, data, date)
     50         display(df_current_day)
     51 
---> 52         self.previous_day = date - 2 * US_BUSINESS_DAY # 2018-7-2
     53         # get the minute data aof previous trading day (2022-04-14)
     54         df_previous_day = df[(df['date'] >= self.previous_day) & (df['date'] <= self.previous_day)]

TypeError: unsupported operand type(s) for -: 'str' and 'pandas._libs.tslibs.offsets.CustomBusinessDay'

How do I fix this?

Update

Types:

<class 'pandas._libs.tslibs.offsets.CustomBusinessDay'>  US_BUSINESS_DAY type
<class 'pandas._libs.tslibs.timestamps.Timestamp'> self.date type

CodePudding user response：

try to put date to this format : datetime64 :

from pandas.tseries.holiday import USFederalHolidayCalendar
from pandas.tseries.offsets import CustomBusinessDay

US_BUSINESS_DAY = CustomBusinessDay(calendar=USFederalHolidayCalendar())


date='2018-7-2'
pd.to_datetime(date) - 2 * US_BUSINESS_DAY # 2018-7-2
Timestamp('2018-06-28 00:00:00')