CODEWITHSUNDEEP
pythonpandasdatetimegroup-bypandas-groupby
Dirk
Dirk

Reputation: 3221

How to group by day and month in pandas?

Given a series like this 

    Date
2005-01-01    128
2005-01-02     72
2005-01-03     67
2005-01-04     61
2005-01-05     33
Name: Data_Value, dtype: int64

for several years, how do I group all the January 1sts together, all the January 2nds, etc?

I'm actually trying to find the max for each day of the year across several years, so it does not have to be groupby. If there is an easier way to do this, that would be great.

Upvotes: 4

Views: 11457

Answers (3)

Vritika Malhotra
Vritika Malhotra

Reputation: 369

Why to not just keep it simple!

max_temp = dfall.groupby([(dfall.Date.dt.month),(dfall.Date.dt.day)])['Data_Value'].max()

Upvotes: 2

Nev1111
Nev1111

Reputation: 1049

As an alternative, you can use a pivot table:

Reset index and format date columns

df=df.reset_index()
df['date']=pd.to_datetime(df['index'])
df['year']=df['date'].dt.year
df['month']=df['date'].dt.month
df['day']=df['date'].dt.day

Pivot over the month and day columns:

df_grouped=df.pivot_table(index=('month','day'),values='Date',aggfunc='max')

Upvotes: 2

cs95
cs95

Reputation: 402293

You can convert your index to datetime, then use strftime to get a date formatted string to group on:

df.groupby(pd.to_datetime(df.index).strftime('%b-%d'))['Date_Value'].max()

If there are no NaNs in your date string, you can slice as well. This returns strings of the format "MM-DD":

df.groupby(df.index.astype(str).str[5:])['Date_Value'].max()

Upvotes: 4

Related Questions