pandas group by week & get last week data as new df

Question

I have a df that has the following, df.dtypes:

key                 object
date                datetime64[ns]
username            object
answer              object
grade               object
dtype: object

I then group by week:

test_lastweek = df.groupby([pd.Grouper(key='date', freq='W-SAT')])['key'].count()

I can see that that are 45 records that fall with in the last week of 2019-08-17:

date
2019-06-29     475
2019-07-06     294
2019-07-13    2311
2019-07-20     389
2019-07-27     554
2019-08-03     408
2019-08-10     587
2019-08-17      45
Freq: W-SAT, Name: key, dtype: int64

Question: How do I get the last weeks data, all 45 records with the data from df and make that into a new df?

Answer 1

I think you can filter the df by the date column as

# Simple dates to datetimes  of from datetime import datetime
df['Date'] = pd.to_datetime(df['Date'])

# Extract the week from the datetime column
df['Week'] = df['Date'].datetime.isocalendar().week

With this code, you can get the week of all your dates and then you'll have to filter it by the week you. Week values are int type so you can do math operations with it, so you can subtract -1 to get a the previous week of the one selected.

max_sem = df['Week'].max()
last_sem = max_sem-1

The example is one I used to get the current week for some data and its predecessor.

Answer 2

you can just filter the dataframe using .loc if you want all records from last week

df_last_week = df.loc[df['date'] >= '2019-08-17']

let me know if this helps.