How to convert YYYYMM to YYYY-MM datetime format without day?-CodePudding

I have two datasets that have monthly frequencies. For one of them,df, I had to aggregate some data to turn it from daily to monthly using the following code:

df_grouped=df.groupby([df.index.to_period('M'),'City ID']).agg({'Estimated Population':'mean','Estimated Population_2019':'mean','Confirmed Rate':['mean','std'],'Death Rate':['mean','std'],'New Confirmed':'sum','New Deaths':'sum'})

df_grouped.index.rename(['Month','City ID'],inplace=True)

After doing these changes my dates became in the format YYYY-MM, for example:

2020-01
2020-02
...
2021-07

My other dataset, df2, has the date in format YYYMM, so I used the following code to convert it:

df2['DATE'] = pd.to_datetime(df2['DATE'],format='%Y%m')

My new dates become in the format YYYY-MM-DD, where all the DD become 01, as follows:

2020-01-01
2020-02-01
...
2021-07-01

How can I convert df2 date now from YYYY-MM-DD to YYYY-MM?

I was thinking, maybe there is a way to convert straight from YYYYMM to YYYY-MM?

CodePudding user response：

You can use

df2['DATE'] = pd.to_datetime(df2['DATE'],format='%Y%m').dt.strftime('%Y-%m')
# or
df['date'] = df['date'].astype(str).str.replace(r'(\d{4})(\d{2})', r'\1-\2', regex=True)

CodePudding user response：

Convert your column to PeriodIndex:

>>> pd.PeriodIndex(df2['DATE'], freq='M')
PeriodIndex(['2020-09', '2020-10', '2020-11', '2020-12', '2021-01', '2021-02'],
            dtype='period[M]', name='DATE')

Input data:

dti = pd.date_range('2020-9', '2021-3', freq='M')
df2 = pd.DataFrame({'DATE': dti.strftime('%Y%m')})
print(df2)

# Output
     DATE
0  202009
1  202010
2  202011
3  202012
4  202101
5  202102

CodePudding user response：

Use:

df = pd.DataFrame({'date': ['2020-01-01', '2020-02-01']})
pd.to_datetime(df['date']).dt.to_period('M')

Output: