Python pandas 将 function 应用于分组 Z6A8064B5DF4794555500553ZD77

Question

I am working on Titanic dataset
One ticket can be issued for several passengers, ie several passengers may have same ticket number
The 'Fare' feature for all those passengers will be the same and equal to the whole ticket fare
For example if there are 4 passengers travelling by one ticket, the ticket fare can be 40$, but each passenger fare should be $10.
So one should divide ticket fare by ticket frequency to calculate fare per passenger.
But there is one more thing: babies are charged $2 and children younger 12 are charged half adult fare.
So I am trying to calculate price payed by each adult in the ticket taking into account children fares.
Here is a sample data frame:

df = pd.DataFrame({'Age': [0.5,5,20,21,22,23,24], 'Fare': [17,17,17,40,40,40,40], 'TicketNum': [1,1,1,2,2,2,2]})

0      0.5       17         1

1      5.0       17         1

2      20.0      17         1

3      21.0      40         2

4      22.0      40         2

5      23.0      40         2

6      24.0      40         2

first I make this function:

def fare_calc(x):
    ticket_fare = x['Fare'].mean()

    group_size = x.shape[0]
    babies_count = x[x['Age']<1].count()
    child_count = x[x['Age']<12].count()
    adult_count = group_size - babies_count - child_count
    adult_fare = (ticket_fare - babies_count * 2) / (adult_count + child_count*0.5)
    return adult_fare

then I try:

df['TicketFreq'] = df.groupby('TicketNum')['TicketNum'].transform('count')
df['Fare2'] = df[df.TicketFreq>1].groupby(['TicketNum'])['Age','Fare'].agg(fare_calc)

and get an error:
ValueError: Wrong number of items passed 2, placement implies 1

desired output is the following:

0      0.5       17         1           10

1      5.0       17         1           10

2      20.0      17         1           10

3      21.0      40         2           10

4      22.0      40         2           10

5      23.0      40         2           10

6      24.0      40         2           10

Answer 1

Hey your formular seems to be wrong however the fare_calc function gets exectued when u replace the .agg call by .apply and remove the two columns you specified. See example below

df[df.TicketFreq>1].groupby(['TicketNum']).apply(fare_calc)

Further there were just a few changes necessary for your function.
To get numeric numvers for the babie_count and child_count you need to specify a colum to get only one integer

def fare_calc(x):
    ticket_fare = x['Fare'].mean()
    group_size = x.shape[0]
    babies_count = x[x['Age']<1]['Age'].count()
    child_count = x[x['Age']<12]['Age'].count()
    adult_count = group_size - babies_count - child_count
    adult_fare = (ticket_fare - babies_count * 2) / (adult_count + child_count * 0.5)
    return adult_fare

Answer 2

Here is my solution

I create columnwise series of values using the pd.Series() and .repeat() functions.

By the way, You forgot to exclude babies_count from child_count using (df['Age']<12) & (df['Age']>1)

def fare_calc(x):
    group_size   = x.shape[0]
    ticket_fare  = pd.Series(x['Fare'].mean().repeat(group_size))
    babies_count = x[x['Age']<1 ]['Age'].count()
    child_count  = x[(df['Age']<12) & (df['Age']>1)]['Age'].count()
    adult_count  = group_size - babies_count - child_count
    adult_fare   = (ticket_fare - babies_count * 2) / (adult_count + child_count * 0.5)
    return adult_fare

And finally extract solely values of the stacked Series created by the apply function using the .values to prevent from "incompatible index" TypeError.

df['Fare2'] = df[df.TicketFreq>1].groupby(['TicketNum']).apply(fare_calc).values

print(df)
    Age  Fare  TicketNum  TicketFreq  Fare2
0   0.5    17          1           3   10.0
1   5.0    17          1           3   10.0
2  20.0    17          1           3   10.0
3  21.0    40          2           4   10.0
4  22.0    40          2           4   10.0
5  23.0    40          2           4   10.0
6  24.0    40          2           4   10.0

EDIT 1: more intuitive version of the previous function:

import pandas as pd

df = pd.DataFrame({'Age': [0.5,5,20,21,22,23,24], 'Fare': [17,17,17,40,40,40,40], 'TicketNum': [1,1,1,2,2,2,2]})
df['TicketFreq'] = df.groupby('TicketNum')['TicketNum'].transform('count')

def fare_calc(x):
    group_size       = x.shape[0]
    x['ticket_fare'] = x['Fare'].mean()
    babies_count     = x[x['Age']<1 ]['Age'].count()
    child_count      = x[(df['Age']<12) & (df['Age']>1)]['Age'].count()
    adult_count      = group_size - babies_count - child_count
    x['adult_fare']  = (x['ticket_fare'] - babies_count * 2) / (adult_count + child_count * 0.5)
    return x['adult_fare']

df['Fare2'] = df[df.TicketFreq>1].groupby(['TicketNum']).apply(fare_calc).values

print(df)
    Age  Fare  TicketNum  TicketFreq  Fare2
0   0.5    17          1           3   10.0
1   5.0    17          1           3   10.0
2  20.0    17          1           3   10.0
3  21.0    40          2           4   10.0
4  22.0    40          2           4   10.0
5  23.0    40          2           4   10.0
6  24.0    40          2           4   10.0

EDIT 2: even simpler where 'Fare2' is directly created inside the function

import pandas as pd

df = pd.DataFrame({'Age': [0.5,5,20,21,22,23,24], 'Fare': [17,17,17,40,40,40,40], 'TicketNum': [1,1,1,2,2,2,2]})
df['TicketFreq'] = df.groupby('TicketNum')['TicketNum'].transform('count')

def fare_calc(x):
    group_size       = x.shape[0]
    ticket_fare      = x['Fare'].mean()
    babies_count     = x[x['Age']<1 ]['Age'].count()
    child_count      = x[(df['Age']<12) & (df['Age']>1)]['Age'].count()
    adult_count      = group_size - babies_count - child_count
    x['Fare2']       = (ticket_fare - babies_count * 2) / (adult_count + child_count * 0.5)
    return x

df = df[df.TicketFreq>1].groupby(['TicketNum']).apply(fare_calc)

print(df)
    Age  Fare  TicketNum  TicketFreq  Fare2
0   0.5    17          1           3   10.0
1   5.0    17          1           3   10.0
2  20.0    17          1           3   10.0
3  21.0    40          2           4   10.0
4  22.0    40          2           4   10.0
5  23.0    40          2           4   10.0
6  24.0    40          2           4   10.0

Answer 3

A minor but immediate problem is that in your last line of code, ['Age', 'Fare'] should be [['Age', 'Fare']] , as you want to index with a list of column names.

The main issue is that you have written fare_calc() to work on the whole DataFrame, but the function that is passed to df.agg() will be applied to each column individually.