[PYTHON] I want to give a group_id to a pandas data frame

Introduction

If you want to delete duplicates with pandas, or if you want to aggregate, you can use drop_duplicates or groupby.

How to remove duplicate elements in Pandas DataFrame or Series Python How to use Pandas groupby

However, there are times when I want to assign a group_id to each group under the same conditions as when groupby, but I didn't know how to do it, so I implemented it. (It may not be a best practice, but it was easy to implement)

Granting group_id

#pandas import
import pandas as pd

#Preparing the data frame
df = pd.DataFrame({
    'building_name': ['Building A', 'Building A', 'B building', 'C building', 'B building', 'B building', 'D building'],
    'property_scale': ['large', 'large', 'small', 'small', 'small', 'small', 'large'],
    'city_code': [1, 1, 1, 2, 1, 1, 1]
})
df

#Group objectization
group_info = df.groupby(['property_scale', 'city_code'])

#Let's take a look at the contents
group_info.groups

{('large', 1): Int64Index([0, 1, 6], dtype='int64'), ('small', 1): Int64Index([2, 4, 5], dtype='int64'), ('small', 2): Int64Index([3], dtype='int64')}

#See also
group_info.get_group(('large', 1))

# group_Granting id
df = pd.concat([
    group_info.get_group(group_name).assign(group_id=group_id)
    for group_id, group_name
    in enumerate(group_info.groups.keys())])
df

I will also make it a function

import pandas as pd
from pandas.core.frame import DataFrame

def add_group_id(df: DataFrame, by: list) -> DataFrame:
    """Group for records with duplicate by values_Give id.

    Args:
        df (DataFrame):Arbitrary data frame
        by (list):Column name to group

    Returns:
        DataFrame

    """
    #Already group_If the id column is included, group in by_Add id as well
    if 'group_id' in df.columns:
        by += ['group_id']
    group_info = df.groupby(by=by)
    new_df = pd.concat([
        group_info.get_group(group_name).assign(group_id=group_id)
        for group_id, group_name
        in enumerate(group_info.groups.keys())])
    return new_df

Postscript

Thanks to @r_beginners for commenting, it seems that groupby has a group_id calculation function in the first place.

import pandas as pd
from pandas.core.frame import DataFrame

def add_group_id(df: DataFrame, by: list) -> DataFrame:
    """Group for records with duplicate by values_Give id.

    Args:
        df (DataFrame):Arbitrary data frame
        by (list):Column name to group

    Returns:
        DataFrame

    """
    #Already group_If the id column is included, group in by_Add id as well
    if 'group_id' in df.columns:
        by += ['group_id']
    new_df = df.assign(group_id =df.groupby(by).ngroup())
    return new_df

As @nkay commented, pd.factorize () seems to work as well.

pandas methods Let's study more. ..