تحلیل خوشه‌بندی مشتریان بانک با K-Means

در این پروژه، داده های مشتریان و کمپین های بازاریابی بانک تحلیل شد تا الگوهای رفتاری مشابه شناسایی شوند. پس از پاک سازی، تبدیل ویژگی های دسته ای، و استانداردسازی داده ها، الگوریتم K-Means برای تقسیم مشتریان به دو خوشه استفاده شد. در ادامه، ارتباط ویژگی های مهم با خوشه ها و کیفیت تقسیم بندی با روش Elbow بررسی شد.

ابزارها

Bank Clustring

In [2]:

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns

In [3]:

df = pd.read_csv('../DATA/bank-full.csv')

In [4]:

df.head()

Out[4]:

	age	job	marital	education	default	housing	loan	contact	month	day_of_week	...	campaign	pdays	poutcome	emp.var.rate	cons.price.idx	cons.conf.idx	euribor3m	nr.employed	subscribed
0	56	housemaid	married	basic.4y	no	no	no	telephone	may	mon	...	1	999	nonexistent	1.1	93.994	-36.4	4.857	5191.0	no
1	57	services	married	high.school	unknown	no	no	telephone	may	mon	...	1	999	nonexistent	1.1	93.994	-36.4	4.857	5191.0	no
2	37	services	married	high.school	no	yes	no	telephone	may	mon	...	1	999	nonexistent	1.1	93.994	-36.4	4.857	5191.0	no
3	40	admin.	married	basic.6y	no	no	no	telephone	may	mon	...	1	999	nonexistent	1.1	93.994	-36.4	4.857	5191.0	no
4	56	services	married	high.school	no	no	yes	telephone	may	mon	...	1	999	nonexistent	1.1	93.994	-36.4	4.857	5191.0	no

5 rows × 21 columns

In [5]:

df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 41188 entries, 0 to 41187
Data columns (total 21 columns):
 #   Column          Non-Null Count  Dtype  
---  ------          --------------  -----  
 0   age             41188 non-null  int64  
 1   job             41188 non-null  object 
 2   marital         41188 non-null  object 
 3   education       41188 non-null  object 
 4   default         41188 non-null  object 
 5   housing         41188 non-null  object 
 6   loan            41188 non-null  object 
 7   contact         41188 non-null  object 
 8   month           41188 non-null  object 
 9   day_of_week     41188 non-null  object 
 10  duration        41188 non-null  int64  
 11  campaign        41188 non-null  int64  
 12  pdays           41188 non-null  int64  
 13  previous        41188 non-null  int64  
 14  poutcome        41188 non-null  object 
 15  emp.var.rate    41188 non-null  float64
 16  cons.price.idx  41188 non-null  float64
 17  cons.conf.idx   41188 non-null  float64
 18  euribor3m       41188 non-null  float64
 19  nr.employed     41188 non-null  float64
 20  subscribed      41188 non-null  object 
dtypes: float64(5), int64(5), object(11)
memory usage: 6.6+ MB

In [6]:

df.describe().transpose()

Out[6]:

	count	mean	std	min	25%	50%	75%	max
age	41188.0	40.024060	10.421250	17.000	32.000	38.000	47.000	98.000
duration	41188.0	258.285010	259.279249	0.000	102.000	180.000	319.000	4918.000
campaign	41188.0	2.567593	2.770014	1.000	1.000	2.000	3.000	56.000
pdays	41188.0	962.475454	186.910907	0.000	999.000	999.000	999.000	999.000
previous	41188.0	0.172963	0.494901	0.000	0.000	0.000	0.000	7.000
emp.var.rate	41188.0	0.081886	1.570960	-3.400	-1.800	1.100	1.400	1.400
cons.price.idx	41188.0	93.575664	0.578840	92.201	93.075	93.749	93.994	94.767
cons.conf.idx	41188.0	-40.502600	4.628198	-50.800	-42.700	-41.800	-36.400	-26.900
euribor3m	41188.0	3.621291	1.734447	0.634	1.344	4.857	4.961	5.045
nr.employed	41188.0	5167.035911	72.251528	4963.600	5099.100	5191.000	5228.100	5228.100

In [10]:

sns.histplot(df, x = 'age', bins = 40,hue = 'loan')

C:\Users\Zahra\anaconda3\Lib\site-packages\seaborn\_oldcore.py:1119: FutureWarning: use_inf_as_na option is deprecated and will be removed in a future version. Convert inf values to NaN before operating instead.
  with pd.option_context('mode.use_inf_as_na', True):
C:\Users\Zahra\anaconda3\Lib\site-packages\seaborn\_oldcore.py:1075: FutureWarning: When grouping with a length-1 list-like, you will need to pass a length-1 tuple to get_group in a future version of pandas. Pass `(name,)` instead of `name` to silence this warning.
  data_subset = grouped_data.get_group(pd_key)
C:\Users\Zahra\anaconda3\Lib\site-packages\seaborn\_oldcore.py:1075: FutureWarning: When grouping with a length-1 list-like, you will need to pass a length-1 tuple to get_group in a future version of pandas. Pass `(name,)` instead of `name` to silence this warning.
  data_subset = grouped_data.get_group(pd_key)
C:\Users\Zahra\anaconda3\Lib\site-packages\seaborn\_oldcore.py:1075: FutureWarning: When grouping with a length-1 list-like, you will need to pass a length-1 tuple to get_group in a future version of pandas. Pass `(name,)` instead of `name` to silence this warning.
  data_subset = grouped_data.get_group(pd_key)
C:\Users\Zahra\anaconda3\Lib\site-packages\seaborn\_oldcore.py:1075: FutureWarning: When grouping with a length-1 list-like, you will need to pass a length-1 tuple to get_group in a future version of pandas. Pass `(name,)` instead of `name` to silence this warning.
  data_subset = grouped_data.get_group(pd_key)

Out[10]:

<Axes: xlabel='age', ylabel='Count'>

In [14]:

plt.figure(figsize=(12,6), dpi = 200)
sns.histplot(data= df[df['pdays']!=999], x = 'pdays')

C:\Users\Zahra\anaconda3\Lib\site-packages\seaborn\_oldcore.py:1119: FutureWarning: use_inf_as_na option is deprecated and will be removed in a future version. Convert inf values to NaN before operating instead.
  with pd.option_context('mode.use_inf_as_na', True):

Out[14]:

<Axes: xlabel='pdays', ylabel='Count'>

In [15]:

df['contact'].unique()

Out[15]:

array(['telephone', 'cellular'], dtype=object)

In [16]:

df['duration']

Out[16]:

0        261
1        149
2        226
3        151
4        307
        ... 
41183    334
41184    383
41185    189
41186    442
41187    239
Name: duration, Length: 41188, dtype: int64

In [19]:

plt.figure(figsize=(12,6), dpi = 200)
sns.histplot(df, x = 'duration', hue = 'contact')
plt.xlim(0,1000)

C:\Users\Zahra\anaconda3\Lib\site-packages\seaborn\_oldcore.py:1119: FutureWarning: use_inf_as_na option is deprecated and will be removed in a future version. Convert inf values to NaN before operating instead.
  with pd.option_context('mode.use_inf_as_na', True):
C:\Users\Zahra\anaconda3\Lib\site-packages\seaborn\_oldcore.py:1075: FutureWarning: When grouping with a length-1 list-like, you will need to pass a length-1 tuple to get_group in a future version of pandas. Pass `(name,)` instead of `name` to silence this warning.
  data_subset = grouped_data.get_group(pd_key)
C:\Users\Zahra\anaconda3\Lib\site-packages\seaborn\_oldcore.py:1075: FutureWarning: When grouping with a length-1 list-like, you will need to pass a length-1 tuple to get_group in a future version of pandas. Pass `(name,)` instead of `name` to silence this warning.
  data_subset = grouped_data.get_group(pd_key)
C:\Users\Zahra\anaconda3\Lib\site-packages\seaborn\_oldcore.py:1075: FutureWarning: When grouping with a length-1 list-like, you will need to pass a length-1 tuple to get_group in a future version of pandas. Pass `(name,)` instead of `name` to silence this warning.
  data_subset = grouped_data.get_group(pd_key)

Out[19]:

(0.0, 1000.0)

In [20]:

df['job'].unique()

Out[20]:

array(['housemaid', 'services', 'admin.', 'blue-collar', 'technician',
       'retired', 'management', 'unemployed', 'self-employed', 'unknown',
       'entrepreneur', 'student'], dtype=object)

In [25]:

df['job'].value_counts().index

Out[25]:

Index(['admin.', 'blue-collar', 'technician', 'services', 'management',
       'retired', 'entrepreneur', 'self-employed', 'housemaid', 'unemployed',
       'student', 'unknown'],
      dtype='object', name='job')

In [26]:

sns.countplot(df, x = 'job', order = df['job'].value_counts().index)
plt.xticks(rotation = 90);

In [28]:

sns.countplot(df, x = 'education', order = df['education'].value_counts().index, hue = "default")
plt.xticks(rotation = 90);

In [30]:

df['default'].value_counts()

Out[30]:

default
no         32588
unknown     8597
yes            3
Name: count, dtype: int64

In [32]:

df["loan"].value_counts()

Out[32]:

loan
no         33950
yes         6248
unknown      990
Name: count, dtype: int64

In [33]:

sns.countplot(df, x = 'default')

Out[33]:

<Axes: xlabel='default', ylabel='count'>

In [34]:

sns.pairplot(df)

C:\Users\Zahra\anaconda3\Lib\site-packages\seaborn\_oldcore.py:1119: FutureWarning: use_inf_as_na option is deprecated and will be removed in a future version. Convert inf values to NaN before operating instead.
with pd.option_context('mode.use_inf_as_na', True):
C:\Users\Zahra\anaconda3\Lib\site-packages\seaborn\_oldcore.py:1119: FutureWarning: use_inf_as_na option is deprecated and will be removed in a future version. Convert inf values to NaN before operating instead.
with pd.option_context('mode.use_inf_as_na', True):
C:\Users\Zahra\anaconda3\Lib\site-packages\seaborn\_oldcore.py:1119: FutureWarning: use_inf_as_na option is deprecated and will be removed in a future version. Convert inf values to NaN before operating instead.
with pd.option_context('mode.use_inf_as_na', True):
C:\Users\Zahra\anaconda3\Lib\site-packages\seaborn\_oldcore.py:1119: FutureWarning: use_inf_as_na option is deprecated and will be removed in a future version. Convert inf values to NaN before operating instead.
with pd.option_context('mode.use_inf_as_na', True):
C:\Users\Zahra\anaconda3\Lib\site-packages\seaborn\_oldcore.py:1119: FutureWarning: use_inf_as_na option is deprecated and will be removed in a future version. Convert inf values to NaN before operating instead.
with pd.option_context('mode.use_inf_as_na', True):
C:\Users\Zahra\anaconda3\Lib\site-packages\seaborn\_oldcore.py:1119: FutureWarning: use_inf_as_na option is deprecated and will be removed in a future version. Convert inf values to NaN before operating instead.
with pd.option_context('mode.use_inf_as_na', True):
C:\Users\Zahra\anaconda3\Lib\site-packages\seaborn\_oldcore.py:1119: FutureWarning: use_inf_as_na option is deprecated and will be removed in a future version. Convert inf values to NaN before operating instead.
with pd.option_context('mode.use_inf_as_na', True):
C:\Users\Zahra\anaconda3\Lib\site-packages\seaborn\_oldcore.py:1119: FutureWarning: use_inf_as_na option is deprecated and will be removed in a future version. Convert inf values to NaN before operating instead.
with pd.option_context('mode.use_inf_as_na', True):
C:\Users\Zahra\anaconda3\Lib\site-packages\seaborn\_oldcore.py:1119: FutureWarning: use_inf_as_na option is deprecated and will be removed in a future version. Convert inf values to NaN before operating instead.
with pd.option_context('mode.use_inf_as_na', True):
C:\Users\Zahra\anaconda3\Lib\site-packages\seaborn\_oldcore.py:1119: FutureWarning: use_inf_as_na option is deprecated and will be removed in a future version. Convert inf values to NaN before operating instead.
with pd.option_context('mode.use_inf_as_na', True):

Out[34]:

<seaborn.axisgrid.PairGrid at 0x1fbf4320c10>

In [37]:

X = pd.get_dummies(df).astype(int)

In [38]:

Out[38]:

	age	duration	campaign	pdays	previous	emp.var.rate	cons.price.idx	cons.conf.idx	euribor3m	nr.employed	...	day_of_week_fri	day_of_week_mon	day_of_week_thu	day_of_week_tue	day_of_week_wed	poutcome_failure	poutcome_nonexistent	poutcome_success	subscribed_no	subscribed_yes
0	56	261	1	999	0	1	93	-36	4	5191	...	0	1	0	0	0	0	1	0	1	0
1	57	149	1	999	0	1	93	-36	4	5191	...	0	1	0	0	0	0	1	0	1	0
2	37	226	1	999	0	1	93	-36	4	5191	...	0	1	0	0	0	0	1	0	1	0
3	40	151	1	999	0	1	93	-36	4	5191	...	0	1	0	0	0	0	1	0	1	0
4	56	307	1	999	0	1	93	-36	4	5191	...	0	1	0	0	0	0	1	0	1	0
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
41183	73	334	1	999	0	-1	94	-50	1	4963	...	1	0	0	0	0	0	1	0	0	1
41184	46	383	1	999	0	-1	94	-50	1	4963	...	1	0	0	0	0	0	1	0	1	0
41185	56	189	2	999	0	-1	94	-50	1	4963	...	1	0	0	0	0	0	1	0	1	0
41186	44	442	1	999	0	-1	94	-50	1	4963	...	1	0	0	0	0	0	1	0	0	1
41187	74	239	3	999	1	-1	94	-50	1	4963	...	1	0	0	0	0	1	0	0	1	0

41188 rows × 65 columns

In [39]:

#we need to scale our data because of the distace method

In [40]:

from sklearn.preprocessing import StandardScaler

In [41]:

scaler = StandardScaler()

In [42]:

scaled_X = scaler.fit_transform(X)

In [43]:

from sklearn.cluster import KMeans

In [44]:

kmeans_model = KMeans(n_clusters=2)

In [47]:

cluster_labels = kmeans_model.fit_predict(scaled_X)

In [48]:

cluster_labels

Out[48]:

array([1, 1, 1, ..., 0, 0, 0])

In [49]:

X['Cluster'] = cluster_labels

In [50]:

Out[50]:

	age	duration	campaign	pdays	previous	emp.var.rate	cons.price.idx	cons.conf.idx	euribor3m	nr.employed	...	day_of_week_mon	day_of_week_thu	day_of_week_tue	day_of_week_wed	poutcome_failure	poutcome_nonexistent	poutcome_success	subscribed_no	subscribed_yes	Cluster
0	56	261	1	999	0	1	93	-36	4	5191	...	1	0	0	0	0	1	0	1	0	1
1	57	149	1	999	0	1	93	-36	4	5191	...	1	0	0	0	0	1	0	1	0	1
2	37	226	1	999	0	1	93	-36	4	5191	...	1	0	0	0	0	1	0	1	0	1
3	40	151	1	999	0	1	93	-36	4	5191	...	1	0	0	0	0	1	0	1	0	1
4	56	307	1	999	0	1	93	-36	4	5191	...	1	0	0	0	0	1	0	1	0	1
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
41183	73	334	1	999	0	-1	94	-50	1	4963	...	0	0	0	0	0	1	0	0	1	0
41184	46	383	1	999	0	-1	94	-50	1	4963	...	0	0	0	0	0	1	0	1	0	0
41185	56	189	2	999	0	-1	94	-50	1	4963	...	0	0	0	0	0	1	0	1	0	0
41186	44	442	1	999	0	-1	94	-50	1	4963	...	0	0	0	0	0	1	0	0	1	0
41187	74	239	3	999	1	-1	94	-50	1	4963	...	0	0	0	0	1	0	0	1	0	0

41188 rows × 66 columns

In [58]:

plt.figure(figsize=(12,6), dpi = 200)

X.corr()['Cluster'].iloc[:-1].sort_values().plot(kind = 'bar')

plt.xticks(rotation = 90);

In [59]:

ssd = []

for k in range(2,10):
    kmean_model = KMeans(n_clusters=k)
    kmean_model.fit(scaled_X)
    
    ssd.append(kmean_model.inertia_) #SSD point -- > cluster center

In [60]:

ssd

Out[60]:

[2479086.128811826,
 2390604.734047414,
 2332846.9701420693,
 2246019.022616529,
 2156303.8257264467,
 2218014.358154257,
 2128418.071949139,
 2068618.0740120136]

In [61]:

plt.plot(range(2,10), ssd, 'o--')

Out[61]:

[<matplotlib.lines.Line2D at 0x1fb89787410>]

In [62]:

pd.Series(ssd)

Out[62]:

0    2.479086e+06
1    2.390605e+06
2    2.332847e+06
3    2.246019e+06
4    2.156304e+06
5    2.218014e+06
6    2.128418e+06
7    2.068618e+06
dtype: float64

In [63]:

pd.Series(ssd).diff()

Out[63]:

0             NaN
1   -88481.394764
2   -57757.763905
3   -86827.947526
4   -89715.196890
5    61710.532428
6   -89596.286205
7   -59799.997937
dtype: float64

In [ ]:

تحلیل خوشه‌بندی مشتریان بانک با K-Means

ابزارها

Python

Pandas

NumPy

Matplotlib

Seaborn

Scikit-learn