Subtract a number column (representing years) from a Date column to create a new date colum-CodePudding

I have column of dates ( Dtype = datetime64[ns]) and a column of integers (Dtype = Int16) in the same dataframe (df). The integers column all have values between 20 and 100.

i want to create a new date column that is the Date column minus the integer columns i.e. df:

RepDate | Age | RevisedDate (To be calc)
1/1/19 | 20 | 1/1/99
4/2/22 | 50 | 4/2/72

so far i've tried:

df['RevisedDate'] = df['RepDate'] - pd.DateOffset(years= df['Age'])

but i get a TypeError: cannot convert the series to <class 'int'>

i've also tried deconstrucint and putting back torgether:

df['YofRevDate'] = pd.DatetimeIndex(df['RepDate']).year - df['Age']
df['MofRevDate'] = pd.DatetimeIndex(df['RepDate']).month
df['DofRevDate'] = pd.DatetimeIndex(df['RepDate']).day

df['RevisedDate'] = date(df['YofRevDate'], df['MofRevDate'], df['DofRevDate'])

but i get the same error on the last line.

your help will be much apricated

CodePudding user response：

To achieve that you have 2 ways:

convert Age to something like timedelta in years (not supported in recent pandas versions), and add timedelta to date column
parse date column, do age calculations, create new date column

I'm not sure how fast is my sollution (second option):

new_year = (df.RepDate.dt.year   df.Age).astype(str)
month_day = df.RepDate.dt.month.astype(str)   "-"   df.RepDate.dt.day.astype(str)
df['NewRepDate'] = pd.to_datetime(new_year   "-"   month_day,format="%Y-%m-%d")
df