Import necessary libraries and sub packages.

  Read the data set.

  Select the feature variables.

  Declare the total number of clusters you want.

  Initialize k-means cluster object.

  Pass feature variable.

  Calculate Silhouette score each number of clusters.

  Print the optimal number of cluster

#import libraries

from sklearn import cluster

import sklearn

from sklearn import preprocessing

import matplotlib.pyplot as plt

data=pd.read_excel(‘/home/soft23/soft23/

Sathish/Pythonfiles/flyer.xlsx’)

#declare empty list to store silhouette_score

silhouette_score_values=list()

#select feature variable

X = data[[‘FlyingReturnsMiles’,’FlightTrans’]]

scaler = preprocessing.StandardScaler()

scaled_df = scaler.fit_transform(X)

print(“After scaling feature variable\n”,scaled_df)

#initialize how many cluster

NumberOfClusters=range(2,10)

#initialize for loop to calculate silhouette_score for each #cluster

for i in NumberOfClusters:

classifier=cluster.KMeans(i,init=’k-means++’, n_init=10, max_iter=300, tol=0.0001, verbose=0, random_state=None, copy_x=True)

classifier.fit(scaled_df)

labels= classifier.predict(scaled_df)

print (“Number Of Clusters:”)

print (i)

print (“Silhouette score value”)

print(sklearn.metrics.silhouette_score

(scaled_df,labels ,metric=’euclidean’, sample_size=

None, random_state=None))

silhouette_score_values.append

(sklearn.metrics.silhouette_score(scaled_df,labels ,

metric=’euclidean’, sample_size=None, random_state=None))

#plot the NumberOfClusters, silhouette_score_values

plt.plot(NumberOfClusters, silhouette_score_values)

plt.title(“Silhouette score values vs Numbers of Clusters “)

plt.show()

#find optimal number of clusters

Optimal_NumberOf_Components=

NumberOfClusters[silhouette_score_values.index

(max(silhouette_score_values))]

print (“Optimal number of clusters is:”,Optimal_NumberOf_Components)