2/24/2020 3.

3_group_by - Jupyter Notebook

Group By: Split, Apply, Combine

Simple aggregations can give you a flavor of your dataset, but often we would prefer to aggregate conditionally
on some label or index: this is implemented in the so-called groupby operation. The name "group by" comes
from a command in the SQL database language, but it is perhaps more illuminative to think of it in the terms
first coined by Hadley Wickham of Rstats fame: split, apply, combine.

In [1]:

import numpy as np
import pandas as pd

localhost:8888/notebooks/Machine Learning/Python/3.3_group_by.ipynb 1/11

2/24/2020 3.3_group_by - Jupyter Notebook

In [8]:

df = [Link]({'key': ['A', 'B', 'C', 'A', 'B', 'C','B'],

'data': range(7)}, columns=['key', 'data'])
df

Out[8]:

key data

0 A 0

1 B 1

2 C 2

3 A 3

4 B 4

5 C 5

6 B 6

In [4]:

[Link]('key')

Out[4]:

<[Link] object at 0x0000010CC3FF41C8>

Notice that what is returned is not a set of DataFrames, but a DataFrameGroupBy object. This object is where
the magic is: you can think of it as a special view of the DataFrame, which is poised to dig into the groups but
does no actual computation until the aggregation is applied.

In [9]:

[Link]('key').sum()

Out[9]:

data

key

A 3

B 11

C 7

In [13]:

for (key, group) in [Link]('key'):

print("{0:20s} shape={1}".format(key, [Link]))

A shape=(2, 2)
B shape=(3, 2)
C shape=(2, 2)

localhost:8888/notebooks/Machine Learning/Python/3.3_group_by.ipynb 2/11

2/24/2020 3.3_group_by - Jupyter Notebook

In [15]:

nyc_flight = pd.read_csv('data/nyc_flights_2013.csv')

In [27]:

nyc_flight.head(3)

Out[27]:

year month day dep_time dep_delay arr_time arr_delay carrier tailnum flight origin

0 2013 1 1 517.0 2.0 830.0 11.0 UA N14228 1545 EWR

1 2013 1 1 533.0 4.0 850.0 20.0 UA N24211 1714 LGA

2 2013 1 1 542.0 2.0 923.0 33.0 AA N619AA 1141 JFK

SELECT carrier, count(*)

FROM df
GROUP BY carrier

In [26]:

n_by_flight = nyc_flight.groupby("carrier")["carrier"].count()
n_by_flight

Out[26]:

carrier
9E 2982
AA 5094
AS 115
B6 8144
DL 7160
EV 8298
F9 107
FL 527
HA 50
MQ 4137
OO 1
UA 8917
US 3152
VX 716
WN 1915
YV 103
Name: carrier, dtype: int64

localhost:8888/notebooks/Machine Learning/Python/3.3_group_by.ipynb 3/11

2/24/2020 3.3_group_by - Jupyter Notebook

In [3]:

data = pd.read_csv('data/phone_data.csv')
[Link](3)

Out[3]:

index date duration item month network network_type

0 0 15/10/14 06:58 34.429 data 2014-11 data data

1 1 15/10/14 06:58 13.000 call 2014-11 Vodafone mobile

2 2 15/10/14 14:46 23.000 call 2014-11 Meteor mobile

In [30]:

[Link]()

<class '[Link]'>
RangeIndex: 830 entries, 0 to 829
Data columns (total 7 columns):
index 830 non-null int64
date 830 non-null object
duration 830 non-null float64
item 830 non-null object
month 830 non-null object
network 830 non-null object
network_type 830 non-null object
dtypes: float64(1), int64(1), object(5)
memory usage: 45.5+ KB

In [4]:

import dateutil

# Convert date from string to date times

#data['date'] = data['date'].apply([Link], dayfirst=True)
data['date'] = pd.to_datetime(data['date'])

In [33]:

[Link]()

<class '[Link]'>
RangeIndex: 830 entries, 0 to 829
Data columns (total 7 columns):
index 830 non-null int64
date 830 non-null datetime64[ns]
duration 830 non-null float64
item 830 non-null object
month 830 non-null object
network 830 non-null object
network_type 830 non-null object
dtypes: datetime64[ns](1), float64(1), int64(1), object(4)
memory usage: 45.5+ KB

1. How many rows the dataset has?

localhost:8888/notebooks/Machine Learning/Python/3.3_group_by.ipynb 4/11

2/24/2020 3.3_group_by - Jupyter Notebook

In [35]:

data['item'].count()

Out[35]:

830

2. What was the longest phone call / data entry?

In [36]:

data['duration'].max()

Out[36]:

10528.0

3. How many seconds of phone calls are recorded in total?

In [7]:

data[data['item']=='call']['duration'].sum()

Out[7]:

92321.0

4. How many entries are there for each month?

In [38]:

data['month'].value_counts()

Out[38]:

2014-11 230
2015-01 205
2014-12 157
2015-02 137
2015-03 101
Name: month, dtype: int64

5. Number of non-null unique network entries

localhost:8888/notebooks/Machine Learning/Python/3.3_group_by.ipynb 5/11

2/24/2020 3.3_group_by - Jupyter Notebook

In [39]:

data['network'].nunique()

Out[39]:

6. Get the first entry for each month

In [40]:

[Link]('month').first()

Out[40]:

index date duration item network network_type

month

2014-11 0 2014-10-15 [Link] 34.429 data data data

2014-12 228 2014-11-13 [Link] 34.429 data data data

2015-01 381 2014-12-13 [Link] 34.429 data data data

2015-02 577 2015-01-13 [Link] 34.429 data data data

2015-03 729 2015-02-12 [Link] 69.000 call landline landline

7. Get the sum of the durations per month

In [41]:

[Link]('month')['duration'].sum()

Out[41]:

month
2014-11 26639.441
2014-12 14641.870
2015-01 18223.299
2015-02 15522.299
2015-03 22750.441
Name: duration, dtype: float64

8. Get the number of dates / entries in each month

localhost:8888/notebooks/Machine Learning/Python/3.3_group_by.ipynb 6/11

2/24/2020 3.3_group_by - Jupyter Notebook

In [42]:

[Link]('month')['date'].count()

Out[42]:

month
2014-11 230
2014-12 157
2015-01 205
2015-02 137
2015-03 101
Name: date, dtype: int64

9. What is the sum of durations, for calls only, to each network

In [43]:

data[data['item'] == 'call'].groupby('network')['duration'].sum()

Out[43]:

network
Meteor 7200.0
Tesco 13828.0
Three 36464.0
Vodafone 14621.0
landline 18433.0
voicemail 1775.0
Name: duration, dtype: float64

10. How many calls, sms, and data entries are in each month?

localhost:8888/notebooks/Machine Learning/Python/3.3_group_by.ipynb 7/11

2/24/2020 3.3_group_by - Jupyter Notebook

In [44]:

[Link](['month', 'item'])['date'].count()

Out[44]:

month item
2014-11 call 107
data 29
sms 94
2014-12 call 79
data 30
sms 48
2015-01 call 88
data 31
sms 86
2015-02 call 67
data 31
sms 39
2015-03 call 47
data 29
sms 25
Name: date, dtype: int64

11. How many calls, texts, and data are sent per month, split by network_type?

localhost:8888/notebooks/Machine Learning/Python/3.3_group_by.ipynb 8/11

2/24/2020 3.3_group_by - Jupyter Notebook

In [45]:

[Link](['month', 'network_type'])['date'].count()

Out[45]:

month network_type
2014-11 data 29
landline 5
mobile 189
special 1
voicemail 6
2014-12 data 30
landline 7
mobile 108
voicemail 8
world 4
2015-01 data 31
landline 11
mobile 160
voicemail 3
2015-02 data 31
landline 8
mobile 90
special 2
voicemail 6
2015-03 data 29
landline 11
mobile 54
voicemail 4
world 3
Name: date, dtype: int64

In [48]:

# produces Pandas Series

#[Link]('month')['duration'].sum()
# Produces Pandas DataFrame
[Link]('month')[['duration']].sum()

Out[48]:

duration

month

2014-11 26639.441

2014-12 14641.870

2015-01 18223.299

2015-02 15522.299

2015-03 22750.441

In [1]:

import pandas as pd

localhost:8888/notebooks/Machine Learning/Python/3.3_group_by.ipynb 9/11

2/24/2020 3.3_group_by - Jupyter Notebook

In [6]:

name = [Link](['ram','shyam','kiran','rishi'])

In [7]:

name

Out[7]:

0 ram
1 shyam
2 kiran
3 rishi
dtype: object

In [9]:

mark1 = [Link]([45, 67, 32, 65])

In [10]:

mark2 = [Link]([77, 34, 72, 55])

In [13]:

df=[Link]({name, mark1, mark2})

---------------------------------------------------------------------------
TypeError Traceback (most recent call last)
<ipython-input-13-f00c768bf164> in <module>
----> 1 df=[Link]({name, mark1, mark2})

~\Anaconda3\lib\site-packages\pandas\core\[Link] in __hash__(self)
1884 raise TypeError(
1885 "{0!r} objects are mutable, thus they cannot be"
-> 1886 " hashed".format(self.__class__.__name__)
1887 )
1888

TypeError: 'Series' objects are mutable, thus they cannot be hashed

In [12]:

df

Out[12]:

0 1 2 3

0 ram shyam kiran rishi

1 45 67 32 65

2 77 34 72 55

localhost:8888/notebooks/Machine Learning/Python/3.3_group_by.ipynb 10/11

2/24/2020 3.3_group_by - Jupyter Notebook

In [ ]:

localhost:8888/notebooks/Machine Learning/Python/3.3_group_by.ipynb 11/11