Create a week of year feature from a pandas datetime column

weeks_from_pandas_datetime.py

import pandas as pd

def create_week_feature(df:pd.DataFrame, date_col='DateTime'):
  """
  Given a dataframe with a column `date_col` of type pd.DateTime
  """
  if type(df[date_col]) != pd.DateTime:
    df[date_col] = pd.to_datetime(df[date_col])
    
  # Use the pd.Series.dt.isocalendar() method to access
  # additional features, like the week in year
  weeks = df[date_col].dt.isocalendar().week
  
  # add in an indicator column `Is_Week` + the week number, 
  # i.e. `Is_Week52`
  df.loc[:, 'WeekInYear'] = ['Is_Week'+str(i) for i in weeks]

  # dummify the features
  week_features = pd.get_dummies(df['WeekInYear'], dtype=int)
  df = pd.concat([df, week_features], axis=1)
  return df
  

When dealing with the pandas pd.DateTime type we sometimes need to get a week feature that specifies the week in the calendar year. The easiest way to do this is to use the pd.Series.dt.isocalendar() method and then it unlocks the .week attribute on the date.

If you're dealing with a fiscal calendar then you'll need to do something different.