EAST WEST AIRLINES OUTPUT

import pandas as pd

import numpy as np

import matplotlib.pylab as plt

East = pd.read_excel("C:\\Users\\Santosh raj\\Desktop\\pyton study material\\hireachical clustering\\

East_West.xlsx")

East.dtypes

Out[5]:

ID# int64

Balance int64

Qual_miles int64

cc1_miles int64

cc2_miles int64

cc3_miles int64

Bonus_miles int64

Bonus_trans int64

Flight_miles_12mo int64

Flight_trans_12 int64

Days_since_enroll int64

Award? int64

dtype: object
East.info()

<class 'pandas.core.frame.DataFrame'>

RangeIndex: 3999 entries, 0 to 3998

Data columns (total 12 columns):

# Column Non-Null Count Dtype

--- ------ -------------- -----

0 ID# 3999 non-null int64

1 Balance 3999 non-null int64

2 Qual_miles 3999 non-null int64

3 cc1_miles 3999 non-null int64

4 cc2_miles 3999 non-null int64

5 cc3_miles 3999 non-null int64

6 Bonus_miles 3999 non-null int64

7 Bonus_trans 3999 non-null int64

8 Flight_miles_12mo 3999 non-null int64

9 Flight_trans_12 3999 non-null int64

10 Days_since_enroll 3999 non-null int64

11 Award? 3999 non-null int64

dtypes: int64(12)

memory usage: 375.0 KB

#To Detect the outliers

import seaborn as sns

sns.boxplot(East.Balance)

C:\Users\Public\conda\lib\site-packages\seaborn\_decorators.py:36: FutureWarning: Pass the following

variable as a keyword arg: x. From version 0.12, the only valid positional argument will be `data`, and
passing other arguments without an explicit keyword will result in an error or misinterpretation.

warnings.warn(

Out[9]: <AxesSubplot:xlabel='Balance'>

#Removing the outilers

IQR = East['Balance'].quantile(0.75) - East['Balance'].quantile(0.25)

lower_limit = East['Balance'].quantile(0.25) - (IQR * 1.5)

upper_limit = East['Balance'].quantile(0.75) + (IQR * 1.5)

#Apply Winsor

from feature_engine.outliers import Winsorizer

winsor = Winsorizer(capping_method='iqr', # choose IQR rule boundaries or gaussian for mean and std

tail='both', # cap left, right or both tails

fold=1.5,

variables=['Balance'])

East_t = winsor.fit_transform(East[['Balance']])

sns.boxplot(East_t.Balance)

C:\Users\Public\conda\lib\site-packages\seaborn\_decorators.py:36: FutureWarning: Pass the following

variable as a keyword arg: x. From version 0.12, the only valid positional argument will be `data`, and
passing other arguments without an explicit keyword will result in an error or misinterpretation.

warnings.warn(

Out[20]: <AxesSubplot:xlabel='Balance'>


#To find outliers for qual miles

sns.boxplot(East.Qual_miles)

C:\Users\Public\conda\lib\site-packages\seaborn\_decorators.py:36: FutureWarning: Pass the following

variable as a keyword arg: x. From version 0.12, the only valid positional argument will be `data`, and
passing other arguments without an explicit keyword will result in an error or misinterpretation.

warnings.warn(

Out[22]: <AxesSubplot:xlabel='Qual_miles'>


#Removing the outilers

IQR = East['Qual_miles'].quantile(0.75) - East['Qual_miles'].quantile(0.25)

lower_limit = East['Qual_miles'].quantile(0.25) - (IQR * 1.5)

upper_limit = East['Qual_miles'].quantile(0.75) + (IQR * 1.5)

#Apply Winsor

from feature_engine.outliers import Winsorizer

winsor = Winsorizer(capping_method='iqr', # choose IQR rule boundaries or gaussian for mean and std

tail='both', # cap left, right or both tails

 fold=1.5,

variables=['Qual_miles'])

East_Q = winsor.fit_transform(East[['Qual_miles']])

sns.boxplot(East_Q.Qual_miles)

C:\Users\Public\conda\lib\site-packages\seaborn\_decorators.py:36: FutureWarning: Pass the following

variable as a keyword arg: x. From version 0.12, the only valid positional argument will be `data`, and
passing other arguments without an explicit keyword will result in an error or misinterpretation.

warnings.warn(

Out[31]: <AxesSubplot:xlabel='Qual_miles'>

#To find outiler for cc1_miles

sns.boxplot(East.cc1_miles)
C:\Users\Public\conda\lib\site-packages\seaborn\_decorators.py:36: FutureWarning: Pass the following
variable as a keyword arg: x. From version 0.12, the only valid positional argument will be `data`, and
passing other arguments without an explicit keyword will result in an error or misinterpretation.

warnings.warn(

Out[33]: <AxesSubplot:xlabel='cc1_miles'>

#Outilers not present

#To find outilers for cc2_miles

sns.boxplot(East.cc2_miles)

C:\Users\Public\conda\lib\site-packages\seaborn\_decorators.py:36: FutureWarning: Pass the following

variable as a keyword arg: x. From version 0.12, the only valid positional argument will be `data`, and
passing other arguments without an explicit keyword will result in an error or misinterpretation.

warnings.warn(

Out[36]: <AxesSubplot:xlabel='cc2_miles'>


#To detect outliers

IQR = East['cc2_miles'].quantile(0.75) - East['cc2_miles'].quantile(0.25)

lower_limit = East['cc2_miles'].quantile(0.25) - (IQR * 1.5)

upper_limit = East['cc2_miles'].quantile(0.75) + (IQR * 1.5)

#Apply Winsor

from feature_engine.outliers import Winsorizer

winsor = Winsorizer(capping_method='iqr', # choose IQR rule boundaries or gaussian for mean and std

tail='both', # cap left, right or both tails

 fold=1.5,

variables=['cc2_miles'])

East_c = winsor.fit_transform(East[['cc2_miles']])

sns.boxplot(East_c.cc2_miles)

C:\Users\Public\conda\lib\site-packages\seaborn\_decorators.py:36: FutureWarning: Pass the following

variable as a keyword arg: x. From version 0.12, the only valid positional argument will be `data`, and
passing other arguments without an explicit keyword will result in an error or misinterpretation.

warnings.warn(

Out[45]: <AxesSubplot:xlabel='cc2_miles'>

#To find outilers for cc3

sns.boxplot(East.cc3_miles)
C:\Users\Public\conda\lib\site-packages\seaborn\_decorators.py:36: FutureWarning: Pass the following
variable as a keyword arg: x. From version 0.12, the only valid positional argument will be `data`, and
passing other arguments without an explicit keyword will result in an error or misinterpretation.

warnings.warn(

Out[47]: <AxesSubplot:xlabel='cc3_miles'>

#To detect outilers

IQR = East['cc3_miles'].quantile(0.75) - East['cc3_miles'].quantile(0.25)

lower_limit = East['cc3_miles'].quantile(0.25) - (IQR * 1.5)

upper_limit = East['cc3_miles'].quantile(0.75) + (IQR * 1.5)

#Apply Winsor
from feature_engine.outliers import Winsorizer

winsor = Winsorizer(capping_method='iqr', # choose IQR rule boundaries or gaussian for mean and std

tail='both', # cap left, right or both tails

fold=1.5,

variables=['cc3_miles'])

East_c3 = winsor.fit_transform(East[['cc3_miles']])

sns.boxplot(East_c3.cc3_miles)

C:\Users\Public\conda\lib\site-packages\seaborn\_decorators.py:36: FutureWarning: Pass the following

variable as a keyword arg: x. From version 0.12, the only valid positional argument will be `data`, and
passing other arguments without an explicit keyword will result in an error or misinterpretation.

warnings.warn(

Out[56]: <AxesSubplot:xlabel='cc3_miles'>


#To find outilers for Bonus_miles

sns.boxplot(East.Bonus_miles)

C:\Users\Public\conda\lib\site-packages\seaborn\_decorators.py:36: FutureWarning: Pass the following

variable as a keyword arg: x. From version 0.12, the only valid positional argument will be `data`, and
passing other arguments without an explicit keyword will result in an error or misinterpretation.

warnings.warn(

Out[58]: <AxesSubplot:xlabel='Bonus_miles'>


#To detect outilers

IQR = East['Bonus_miles'].quantile(0.75) - East['Bonus_miles'].quantile(0.25)

lower_limit = East['Bonus_miles'].quantile(0.25) - (IQR * 1.5)

upper_limit = East['Bonus_miles'].quantile(0.75) + (IQR * 1.5)

#Apply Winsor

from feature_engine.outliers import Winsorizer

winsor = Winsorizer(capping_method='iqr', # choose IQR rule boundaries or gaussian for mean and std

tail='both', # cap left, right or both tails

 fold=1.5,

variables=['Bonus_miles'])

winsor = Winsorizer(capping_method='iqr', # choose IQR rule boundaries or gaussian for mean and std

tail='both', # cap left, right or both tails

fold=1.5,

variables=['Bonus_miles'])

East_B = winsor.fit_transform(East[['Bonus_miles']])

sns.boxplot(East_B.Bonus_miles)

C:\Users\Public\conda\lib\site-packages\seaborn\_decorators.py:36: FutureWarning: Pass the following

variable as a keyword arg: x. From version 0.12, the only valid positional argument will be `data`, and
passing other arguments without an explicit keyword will result in an error or misinterpretation.

warnings.warn(

Out[68]: <AxesSubplot:xlabel='Bonus_miles'>


#To find outilers for Bonus_tans

sns.boxplot(East.Bonus_trans)

C:\Users\Public\conda\lib\site-packages\seaborn\_decorators.py:36: FutureWarning: Pass the following

variable as a keyword arg: x. From version 0.12, the only valid positional argument will be `data`, and
passing other arguments without an explicit keyword will result in an error or misinterpretation.

warnings.warn(

Out[70]: <AxesSubplot:xlabel='Bonus_trans'>


#To detect outilers

IQR = East['Bonus_trans'].quantile(0.75) - East['Bonus_trans'].quantile(0.25)

lower_limit = East['Bonus_trans'].quantile(0.25) - (IQR * 1.5)

upper_limit = East['Bonus_trans'].quantile(0.75) + (IQR * 1.5)

#Apply Winsor

from feature_engine.outliers import Winsorizer

winsor = Winsorizer(capping_method='iqr', # choose IQR rule boundaries or gaussian for mean and std

tail='both', # cap left, right or both tails

 fold=1.5,

variables=['Bonus_trans'])

East_t = winsor.fit_transform(East[['Bonus_trans']])

sns.boxplot(East_t.Bonus_trans)

C:\Users\Public\conda\lib\site-packages\seaborn\_decorators.py:36: FutureWarning: Pass the following

variable as a keyword arg: x. From version 0.12, the only valid positional argument will be `data`, and
passing other arguments without an explicit keyword will result in an error or misinterpretation.

warnings.warn(

Out[79]: <AxesSubplot:xlabel='Bonus_trans'>

#To find outilers for Flight

sns.boxplot(East.Flight_miles_12mo)
C:\Users\Public\conda\lib\site-packages\seaborn\_decorators.py:36: FutureWarning: Pass the following
variable as a keyword arg: x. From version 0.12, the only valid positional argument will be `data`, and
passing other arguments without an explicit keyword will result in an error or misinterpretation.

warnings.warn(

Out[81]: <AxesSubplot:xlabel='Flight_miles_12mo'>

#To detect outilers

IQR = East['Flight_miles_12mo'].quantile(0.75) - East['Flight_miles_12mo'].quantile(0.25)

lower_limit = East['Flight_miles_12mo'].quantile(0.25) - (IQR * 1.5)

upper_limit = East['Flight_miles_12mo'].quantile(0.75) + (IQR * 1.5)

#Apply Winsor
from feature_engine.outliers import Winsorizer

winsor = Winsorizer(capping_method='iqr', # choose IQR rule boundaries or gaussian for mean and std

tail='both', # cap left, right or both tails

fold=1.5,

variables=['Flight_miles_12mo'])

East_F = winsor.fit_transform(East[['Flight_miles_12mo']])

sns.boxplot(East_F.Flight_miles_12mo)

C:\Users\Public\conda\lib\site-packages\seaborn\_decorators.py:36: FutureWarning: Pass the following

variable as a keyword arg: x. From version 0.12, the only valid positional argument will be `data`, and
passing other arguments without an explicit keyword will result in an error or misinterpretation.

warnings.warn(

Out[90]: <AxesSubplot:xlabel='Flight_miles_12mo'>


#To find Outilers for Flight trans

sns.boxplot(East.Flight_trans_12)

C:\Users\Public\conda\lib\site-packages\seaborn\_decorators.py:36: FutureWarning: Pass the following

variable as a keyword arg: x. From version 0.12, the only valid positional argument will be `data`, and
passing other arguments without an explicit keyword will result in an error or misinterpretation.

warnings.warn(

Out[92]: <AxesSubplot:xlabel='Flight_trans_12'>


#To detect outilers

IQR = East['Flight_trans_12'].quantile(0.75) - East['Flight_trans_12'].quantile(0.25)

lower_limit = East['Flight_trans_12'].quantile(0.25) - (IQR * 1.5)

upper_limit = East['Flight_trans_12'].quantile(0.75) + (IQR * 1.5)

#Apply Winsor

from feature_engine.outliers import Winsorizer

winsor = Winsorizer(capping_method='iqr', # choose IQR rule boundaries or gaussian for mean and std

tail='both', # cap left, right or both tails

 fold=1.5,

variables=['Flight_trans_12'])

East_F2 = winsor.fit_transform(East[['Flight_trans_12']])

sns.boxplot(East_F2.Flight_trans_12)

C:\Users\Public\conda\lib\site-packages\seaborn\_decorators.py:36: FutureWarning: Pass the following

variable as a keyword arg: x. From version 0.12, the only valid positional argument will be `data`, and
passing other arguments without an explicit keyword will result in an error or misinterpretation.

warnings.warn(

Out[101]: <AxesSubplot:xlabel='Flight_trans_12'>

#To find outilers for day since enroll

sns.boxplot(East.Days_since_enroll)
C:\Users\Public\conda\lib\site-packages\seaborn\_decorators.py:36: FutureWarning: Pass the following
variable as a keyword arg: x. From version 0.12, the only valid positional argument will be `data`, and
passing other arguments without an explicit keyword will result in an error or misinterpretation.

warnings.warn(

Out[103]: <AxesSubplot:xlabel='Days_since_enroll'>

#No outlers present

#To find Na Values

East.isna().sum()

Out[106]:

ID# 0

Balance 0

Qual_miles 0
cc1_miles 0

cc2_miles 0

cc3_miles 0

Bonus_miles 0

Bonus_trans 0

Flight_miles_12mo 0

Flight_trans_12 0

Days_since_enroll 0

Award? 0

dtype: int64

#Plot Histogram

plt.hist(East_B.Bonus_miles)

Out[108]:

(array([1855., 490., 340., 269., 209., 151., 142., 113., 84.,

346.]),

array([ 0. , 5762.625, 11525.25 , 17287.875, 23050.5 , 28813.125,

34575.75 , 40338.375, 46101. , 51863.625, 57626.25 ]),

<BarContainer object of 10 artists>)



plt.hist(East_t.Bonus_trans)

Out[109]:

(array([1041., 519., 396., 853., 454., 314., 151., 100., 70.,

101.]),

array([ 0. , 3.8, 7.6, 11.4, 15.2, 19. , 22.8, 26.6, 30.4, 34.2, 38. ]),

<BarContainer object of 10 artists>)



plt.hist(East_F.Flight_miles_12mo)

Out[110]:

(array([2763., 105., 52., 79., 51., 51., 197., 53., 31.,

617.]),

array([ 0. , 77.75, 155.5 , 233.25, 311. , 388.75, 466.5 , 544.25,

622. , 699.75, 777.5 ]),

<BarContainer object of 10 artists>)



plt.hist(East_F2.Flight_trans_12)

Out[111]:

(array([2723., 0., 0., 0., 469., 0., 0., 0., 242.,

565.]),

array([0. , 0.25, 0.5 , 0.75, 1. , 1.25, 1.5 , 1.75, 2. , 2.25, 2.5 ]),

<BarContainer object of 10 artists>)



East1 = East.drop(["ID#"], axis=1)

# Normalization function

def norm_func(i):

x = (i-i.min()) / (i.max()-i.min())

return (x)

# Normalized data frame (considering the numerical part of data)

df_norm = norm_func(East1.iloc[:, 1:])

df_norm.describe()

Out[117]:
 Qual_miles cc1_miles ... Days_since_enroll Award?

count 3999.000000 3999.000000 ... 3999.000000 3999.000000

mean 0.012927 0.264879 ... 0.496330 0.370343

std 0.069399 0.344230 ... 0.248991 0.482957

min 0.000000 0.000000 ... 0.000000 0.000000

25% 0.000000 0.000000 ... 0.280685 0.000000

50% 0.000000 0.000000 ... 0.493610 0.000000

75% 0.000000 0.500000 ... 0.697914 1.000000

max 1.000000 1.000000 ... 1.000000 1.000000

[8 rows x 10 columns]

# for creating dendrogram

from scipy.cluster.hierarchy import linkage, dendrogram

# import scipy.cluster.hierarchy as sch

z = linkage(df_norm, method = "complete", metric = "euclidean")

# Dendrogram

plt.figure(figsize=(15, 8));plt.title('Hierarchical Clustering

Dendrogram');plt.xlabel('Index');plt.ylabel('Distance')

dendrogram(z,

leaf_rotation = 0, # rotates the x axis labels

leaf_font_size = 10 # font size for the x axis labels

)
plt.show()

# Now applying AgglomerativeClustering choosing 5 as clusters from the above dendrogram

from sklearn.cluster import AgglomerativeClustering

h_complete = AgglomerativeClustering(n_clusters = 3, linkage = 'complete', affinity =

"euclidean").fit(df_norm)

h_complete.labels_

Out[125]: array([0, 0, 0, ..., 2, 0, 0], dtype=int64)

cluster_labels = pd.Series(h_complete.labels_)

East1['clust'] = cluster_labels # creating a new column and assigning it to new column

East1.head()

Out[128]:

Balance Qual_miles cc1_miles ... Days_since_enroll Award? clust

0 28143 0 1 ... 7000 0 0

1 19244 0 1 ... 6968 0 0

2 41354 0 1 ... 7034 0 0

3 14776 0 1 ... 6952 0 0

4 97752 0 4 ... 6935 1 2

[5 rows x 12 columns]

# Aggregate mean of each cluster

East1.iloc[:, 1:].groupby(East1.clust).mean()

Out[130]:

Qual_miles cc1_miles cc2_miles ... Days_since_enroll Award? clust

clust ...

0 87.790412 1.703645 1.018621 ... 3822.941363 0.002377 0.0

1 347.000000 2.500000 1.000000 ... 2200.250000 1.000000 1.0

2 240.205982 2.668933 1.007478 ... 4631.008838 1.000000 2.0

[3 rows x 11 columns]
East1.iloc[:, 2:].groupby(East1.clust).std()

Out[131]:

cc1_miles cc2_miles cc3_miles ... Days_since_enroll Award? clust

clust ...

0 1.152351 0.173702 0.156518 ... 2079.882670 0.048708 0.0

1 1.732051 0.000000 0.000000 ... 1171.167331 0.000000 0.0

2 1.510456 0.086180 0.248128 ... 1936.161303 0.000000 0.0

[3 rows x 10 columns]

# creating a csv file

East1.to_csv("East_West.csv", encoding = "utf-8")

import os

os.getcwd()

#In Cluster 1 flying patterns, earning and use of frequent flyer rewards, and use of the airline credit card
usage is high

#In Cluster 2 flying patterns, earning and use of frequent flyer rewards, and use of the airline credit card
usage is medium

#In Cluster 0 flying patterns, earning and use of frequent flyer rewards, and use of the airline credit card
usage is low