pandas dataframe group by create a new column - pandas

Question

I have a dataframe with the below format:
name date
Anne 2018/07/04
Anne 2018/07/06
Bob 2015/10/01
Bob 2015/10/10
Bob 2015/11/11
Anne 2018/07/05
... ...
I would like to add a column which is a relative number of days passed from the minimum date of the person.
for each row:
relative_day = (person's date) - (minimum of person's date)
The output is:
name date relative_day
Anne 2018/07/04 0
Anne 2018/07/04 2
Bob 2015/10/01 0
Bob 2015/10/01 9
Bob 2015/11/11 41
Anne 2018/07/05 1
I tried to groupby name first and then writing a for loop over each name and add a column but it gives the error of
A value is trying to be set on a copy of a slice from a DataFrame.
Here is the code I have tried so far:
df['relative_day'] = None
person_groups = df.groupby('name')
for person_name, person_dates in person_groups:
person_dates['relative_day'] = person_dates['date'].min()

Answer 1

Get the name as an index, group on the name, then subtract the minimum to get your relative dates.
result = df.astype({"date": np.datetime64}).set_index("name")
result.assign(relative_day=result['date'] - result.groupby("name")['date'].transform("min"))
date relative_day
name
Anne 2018-07-04 0 days
Anne 2018-07-06 2 days
Bob 2015-10-01 0 days
Bob 2015-10-10 9 days
Bob 2015-11-11 41 days
Anne 2018-07-05 1 days

Answer 2

Let us try
df.date=pd.to_datetime(df.date)
df['new'] = (df.date - df.groupby('name').date.transform('min')).dt.days
df
name date new
0 Anne 2018-07-04 0
1 Anne 2018-07-06 2
2 Bob 2015-10-01 0
3 Bob 2015-10-10 9
4 Bob 2015-11-11 41
5 Anne 2018-07-05 1

Answer 3

#sammywemmy has a good solution. I want to show another possible way.
import pandas as pd
# read dataset
df = pd.read_csv('data.csv')
# change column data type
df['date'] = pd.to_datetime(df['date'], format='%Y/%m/%d')
# group by name
df_group = df.groupby('name')
# get minimum date value
df_group_min = df_group['date'].min()
# create minimum date column by name
df['min'] = df.apply(lambda r: df_group_min[r['name']], axis=1)
# calculate relative day
df['relative_day'] = (df['date'] - df['min']).dt.days
# remove minimum column
df.drop('min', axis=1, inplace=True)
print(df)
Output
name date relative_day
0 Anne 2018-07-04 0
1 Anne 2018-07-06 2
2 Bob 2015-10-01 0
3 Bob 2015-10-10 9
4 Bob 2015-11-11 41
5 Anne 2018-07-05 1