Machine [unsupervised{Clustering(K-means)}] (평할/분할 기반의 군집)

In [1]:

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
# 비슷한 성향끼리 묶는작업

In [4]:

df = pd.read_csv("../data/Mall_Customers.csv")

In [5]:

df.head()

Out[5]:

	CustomerID	Genre	Age	Annual Income (k$)	Spending Score (1-100)
0	1	Male	19	15	39
1	2	Male	21	15	81
2	3	Female	20	16	6
3	4	Female	23	16	77
4	5	Female	31	17	40

In [6]:

df.isna().sum()

Out[6]:

CustomerID                0
Genre                     0
Age                       0
Annual Income (k$)        0
Spending Score (1-100)    0
dtype: int64

In [7]:

df.describe()  

Out[7]:

	CustomerID	Age	Annual Income (k$)	Spending Score (1-100)
count	200.000000	200.000000	200.000000	200.000000
mean	100.500000	38.850000	60.560000	50.200000
std	57.879185	13.969007	26.264721	25.823522
min	1.000000	18.000000	15.000000	1.000000
25%	50.750000	28.750000	41.500000	34.750000
50%	100.500000	36.000000	61.500000	50.000000
75%	150.250000	49.000000	78.000000	73.000000
max	200.000000	70.000000	137.000000	99.000000

In [8]:

# X 만 셋팅, 왜냐?? 언수퍼바이즈드 러닝이니까, 당연히 y는 없다.

In [9]:

df.head(3)

Out[9]:

	CustomerID	Genre	Age	Annual Income (k$)	Spending Score (1-100)
0	1	Male	19	15	39
1	2	Male	21	15	81
2	3	Female	20	16	6

In [10]:

X = df.loc[ :, "Annual Income (k$)" : ]

In [11]:

X

Out[11]:

	Annual Income (k$)	Spending Score (1-100)
0	15	39
1	15	81
2	16	6
3	16	77
4	17	40
...	...	...
195	120	79
196	126	28
197	126	74
198	137	18
199	137	83

200 rows × 2 columns

In [12]:

# 모델링 

In [13]:

from sklearn.cluster import KMeans

In [14]:

kmeans = KMeans(n_clusters= 3, random_state=2) # 그룹의 갯수 n_clusters=

In [15]:

y_pred = kmeans.fit_predict(X)

C:\Users\5-10\Anaconda3\envs\YH\lib\site-packages\sklearn\cluster\_kmeans.py:1334: UserWarning: KMeans is known to have a memory leak on Windows with MKL, when there are less chunks than available threads. You can avoid it by setting the environment variable OMP_NUM_THREADS=1.
  warnings.warn(

In [16]:

y_pred

Out[16]:

array([0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
       0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
       0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
       0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
       0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
       0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 2, 1, 2, 1, 2, 1, 2, 1,
       2, 1, 2, 1, 2, 1, 2, 1, 2, 1, 2, 1, 2, 1, 2, 1, 2, 1, 2, 1, 2, 1,
       2, 1, 2, 1, 2, 1, 2, 1, 2, 1, 2, 1, 2, 1, 2, 1, 2, 1, 2, 1, 2, 1,
       2, 1, 2, 1, 2, 1, 2, 1, 2, 1, 2, 1, 2, 1, 2, 1, 2, 1, 2, 1, 2, 1,
       2, 1])

In [17]:

df["Group"] = y_pred

In [18]:

df

Out[18]:

	CustomerID	Genre	Age	Annual Income (k$)	Spending Score (1-100)	Group
0	1	Male	19	15	39	0
1	2	Male	21	15	81	0
2	3	Female	20	16	6	0
3	4	Female	23	16	77	0
4	5	Female	31	17	40	0
...	...	...	...	...	...	...
195	196	Female	35	120	79	1
196	197	Female	45	126	28	2
197	198	Male	32	126	74	1
198	199	Male	32	137	18	2
199	200	Male	30	137	83	1

200 rows × 6 columns

In [19]:

# 그룹번호 1번인사람만 가져와라.

In [20]:

df.loc[df["Group"] ==1 , ] 

Out[20]:

	CustomerID	Genre	Age	Annual Income (k$)	Spending Score (1-100)	Group
123	124	Male	39	69	91	1
125	126	Female	31	70	77	1
127	128	Male	40	71	95	1
129	130	Male	38	71	75	1
131	132	Male	39	71	75	1
133	134	Female	31	72	71	1
135	136	Female	29	73	88	1
137	138	Male	32	73	73	1
139	140	Female	35	74	72	1
141	142	Male	32	75	93	1
143	144	Female	32	76	87	1
145	146	Male	28	77	97	1
147	148	Female	32	77	74	1
149	150	Male	34	78	90	1
151	152	Male	39	78	88	1
153	154	Female	38	78	76	1
155	156	Female	27	78	89	1
157	158	Female	30	78	78	1
159	160	Female	30	78	73	1
161	162	Female	29	79	83	1
163	164	Female	31	81	93	1
165	166	Female	36	85	75	1
167	168	Female	33	86	95	1
169	170	Male	32	87	63	1
171	172	Male	28	87	75	1
173	174	Male	36	87	92	1
175	176	Female	30	88	86	1
177	178	Male	27	88	69	1
179	180	Male	35	93	90	1
181	182	Female	32	97	86	1
183	184	Female	29	98	88	1
185	186	Male	30	99	97	1
187	188	Male	28	101	68	1
189	190	Female	36	103	85	1
191	192	Female	32	103	69	1
193	194	Female	38	113	91	1
195	196	Female	35	120	79	1
197	198	Male	32	126	74	1
199	200	Male	30	137	83	1

In [ ]:

In [21]:

# 마케팅 현업에서 자주 사용하는 방법 WCSS
# 그룹의 최적화된 갯수를 구하는방법이다.

In [24]:

wcss = []
for k in np.arange(1, 10+1) :
    kmeans = KMeans(n_clusters = k , random_state=5)
    kmeans.fit(X)
    wcss.append( kmeans.inertia_)

C:\Users\5-10\Anaconda3\envs\YH\lib\site-packages\sklearn\cluster\_kmeans.py:1334: UserWarning: KMeans is known to have a memory leak on Windows with MKL, when there are less chunks than available threads. You can avoid it by setting the environment variable OMP_NUM_THREADS=1.
  warnings.warn(
C:\Users\5-10\Anaconda3\envs\YH\lib\site-packages\sklearn\cluster\_kmeans.py:1334: UserWarning: KMeans is known to have a memory leak on Windows with MKL, when there are less chunks than available threads. You can avoid it by setting the environment variable OMP_NUM_THREADS=1.
  warnings.warn(
C:\Users\5-10\Anaconda3\envs\YH\lib\site-packages\sklearn\cluster\_kmeans.py:1334: UserWarning: KMeans is known to have a memory leak on Windows with MKL, when there are less chunks than available threads. You can avoid it by setting the environment variable OMP_NUM_THREADS=1.
  warnings.warn(
C:\Users\5-10\Anaconda3\envs\YH\lib\site-packages\sklearn\cluster\_kmeans.py:1334: UserWarning: KMeans is known to have a memory leak on Windows with MKL, when there are less chunks than available threads. You can avoid it by setting the environment variable OMP_NUM_THREADS=1.
  warnings.warn(
C:\Users\5-10\Anaconda3\envs\YH\lib\site-packages\sklearn\cluster\_kmeans.py:1334: UserWarning: KMeans is known to have a memory leak on Windows with MKL, when there are less chunks than available threads. You can avoid it by setting the environment variable OMP_NUM_THREADS=1.
  warnings.warn(
C:\Users\5-10\Anaconda3\envs\YH\lib\site-packages\sklearn\cluster\_kmeans.py:1334: UserWarning: KMeans is known to have a memory leak on Windows with MKL, when there are less chunks than available threads. You can avoid it by setting the environment variable OMP_NUM_THREADS=1.
  warnings.warn(
C:\Users\5-10\Anaconda3\envs\YH\lib\site-packages\sklearn\cluster\_kmeans.py:1334: UserWarning: KMeans is known to have a memory leak on Windows with MKL, when there are less chunks than available threads. You can avoid it by setting the environment variable OMP_NUM_THREADS=1.
  warnings.warn(
C:\Users\5-10\Anaconda3\envs\YH\lib\site-packages\sklearn\cluster\_kmeans.py:1334: UserWarning: KMeans is known to have a memory leak on Windows with MKL, when there are less chunks than available threads. You can avoid it by setting the environment variable OMP_NUM_THREADS=1.
  warnings.warn(
C:\Users\5-10\Anaconda3\envs\YH\lib\site-packages\sklearn\cluster\_kmeans.py:1334: UserWarning: KMeans is known to have a memory leak on Windows with MKL, when there are less chunks than available threads. You can avoid it by setting the environment variable OMP_NUM_THREADS=1.
  warnings.warn(
C:\Users\5-10\Anaconda3\envs\YH\lib\site-packages\sklearn\cluster\_kmeans.py:1334: UserWarning: KMeans is known to have a memory leak on Windows with MKL, when there are less chunks than available threads. You can avoid it by setting the environment variable OMP_NUM_THREADS=1.
  warnings.warn(

In [25]:

wcss

Out[25]:

[269981.28,
 181363.59595959596,
 106348.37306211119,
 73679.78903948834,
 44448.45544793371,
 37233.81451071001,
 30259.65720728547,
 25079.7669621159,
 21850.165282585636,
 19712.851860217077]

In [ ]:

# 위의 각 클러스터의 갯수마다 구현
# wcss 값을, 차트로 나타낸다 => 엘보우 메소드라고 한다.

In [27]:

In [29]:

x = np.arange(1, 10+1)
plt.plot(x, wcss)
plt.title("The Elbow Method")
plt.xlabel('Number of Clusters')
plt.ylabel("WCSS")
plt.show()

In [30]:

kmeans = KMeans(n_clusters = 5 , random_state =5 )

In [32]:

y_pred = kmeans.fit_predict(X)

C:\Users\5-10\Anaconda3\envs\YH\lib\site-packages\sklearn\cluster\_kmeans.py:1334: UserWarning: KMeans is known to have a memory leak on Windows with MKL, when there are less chunks than available threads. You can avoid it by setting the environment variable OMP_NUM_THREADS=1.
  warnings.warn(

In [33]:

y_pred

Out[33]:

array([3, 0, 3, 0, 3, 0, 3, 0, 3, 0, 3, 0, 3, 0, 3, 0, 3, 0, 3, 0, 3, 0,
       3, 0, 3, 0, 3, 0, 3, 0, 3, 0, 3, 0, 3, 0, 3, 0, 3, 0, 3, 0, 3, 1,
       3, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 4, 2, 1, 2, 4, 2, 4, 2,
       1, 2, 4, 2, 4, 2, 4, 2, 4, 2, 1, 2, 4, 2, 4, 2, 4, 2, 4, 2, 4, 2,
       4, 2, 4, 2, 4, 2, 4, 2, 4, 2, 4, 2, 4, 2, 4, 2, 4, 2, 4, 2, 4, 2,
       4, 2, 4, 2, 4, 2, 4, 2, 4, 2, 4, 2, 4, 2, 4, 2, 4, 2, 4, 2, 4, 2,
       4, 2])

In [34]:

df["Group"] = y_pred 

In [35]:

df

Out[35]:

	CustomerID	Genre	Age	Annual Income (k$)	Spending Score (1-100)	Group
0	1	Male	19	15	39	3
1	2	Male	21	15	81	0
2	3	Female	20	16	6	3
3	4	Female	23	16	77	0
4	5	Female	31	17	40	3
...	...	...	...	...	...	...
195	196	Female	35	120	79	2
196	197	Female	45	126	28	4
197	198	Male	32	126	74	2
198	199	Male	32	137	18	4
199	200	Male	30	137	83	2

200 rows × 6 columns

In [ ]:

In [37]:

import seaborn as sb

In [39]:

sb.scatterplot(data = df , x = 'Annual Income (k$)', y = 'Spending Score (1-100)')
plt.show()

In [ ]:

In [40]:

plt.figure(figsize=[12,8])
plt.scatter(X.values[y_pred == 0, 0], X.values[y_pred == 0, 1], s = 100, c = 'red', label = 'Cluster 1')
plt.scatter(X.values[y_pred == 1, 0], X.values[y_pred == 1, 1], s = 100, c = 'blue', label = 'Cluster 2')
plt.scatter(X.values[y_pred == 2, 0], X.values[y_pred == 2, 1], s = 100, c = 'green', label = 'Cluster 3')
plt.scatter(X.values[y_pred == 3, 0], X.values[y_pred == 3, 1], s = 100, c = 'cyan', label = 'Cluster 4')
plt.scatter(X.values[y_pred == 4, 0], X.values[y_pred == 4, 1], s = 100, c = 'magenta', label = 'Cluster 5')
plt.scatter(kmeans.cluster_centers_[:, 0], kmeans.cluster_centers_[:, 1], s = 300, c = 'yellow', label = 'Centroids')
plt.title('Clusters of customers')
plt.xlabel('Annual Income (k$)')
plt.ylabel('Spending Score (1-100)')
plt.legend()
plt.show()

In [ ]:

In [ ]:

# 1. 그룹이 3인 사람들의 데이터를 가져오세요.

In [42]:

df.head(1)

Out[42]:

	CustomerID	Genre	Age	Annual Income (k$)	Spending Score (1-100)	Group
0	1	Male	19	15	39	3

In [44]:

df.loc[df["Group"] == 3 , ]

Out[44]:

	CustomerID	Genre	Age	Annual Income (k$)	Spending Score (1-100)	Group
0	1	Male	19	15	39	3
2	3	Female	20	16	6	3
4	5	Female	31	17	40	3
6	7	Female	35	18	6	3
8	9	Male	64	19	3	3
10	11	Male	67	19	14	3
12	13	Female	58	20	15	3
14	15	Male	37	20	13	3
16	17	Female	35	21	35	3
18	19	Male	52	23	29	3
20	21	Male	35	24	35	3
22	23	Female	46	25	5	3
24	25	Female	54	28	14	3
26	27	Female	45	28	32	3
28	29	Female	40	29	31	3
30	31	Male	60	30	4	3
32	33	Male	53	33	4	3
34	35	Female	49	33	14	3
36	37	Female	42	34	17	3
38	39	Female	36	37	26	3
40	41	Female	65	38	35	3
42	43	Male	48	39	36	3
44	45	Female	49	39	28	3

In [49]:

df_group_3 = df.loc[df["Group"] == 3 , ]

In [41]:

# 2. 그룹이 3인 사람들의 수입 평균은 얼마입니까?

In [50]:

df_group_3["Annual Income (k$)"].mean()

Out[50]:

26.304347826086957

In [46]:

df.groupby("Group")["Annual Income (k$)"].mean()

Out[46]:

Group
0    25.727273
1    55.296296
2    86.538462
3    26.304348
4    88.200000
Name: Annual Income (k$), dtype: float64

In [ ]:

# 3. 그룹별 Spending Score 를 나타내되
#    평균값과 최대값을 한번에 보여주세요.

In [47]:

df.groupby("Group")["Spending Score (1-100)"].describe()

Out[47]:

	count	mean	std	min	25%	50%	75%	max
Group
0	22.0	79.363636	10.504174	61.0	73.0	77.0	85.75	99.0
1	81.0	49.518519	6.530909	34.0	44.0	50.0	55.00	61.0
2	39.0	82.128205	9.364489	63.0	74.5	83.0	90.00	97.0
3	23.0	20.913043	13.017167	3.0	9.5	17.0	33.50	40.0
4	35.0	17.114286	9.952154	1.0	10.0	16.0	23.50	39.0

In [51]:

df.groupby("Group")["Spending Score (1-100)"].agg( [ np.mean, np.max])

Out[51]:

	mean	amax
Group
0	79.363636	99
1	49.518519	61
2	82.128205	97
3	20.913043	40
4	17.114286	39

In [52]:

# 각 그룹별, 수입과 소비지표의 평균을 구하시오.

In [53]:

df.head(1)

Out[53]:

	CustomerID	Genre	Age	Annual Income (k$)	Spending Score (1-100)	Group
0	1	Male	19	15	39	3

In [55]:

df.groupby("Group")[['Annual Income (k$)','Spending Score (1-100)']].describe()

Out[55]:

	Annual Income (k$)								Spending Score (1-100)
	count	mean	std	min	25%	50%	75%	max	count	mean	std	min	25%	50%	75%	max
Group
0	22.0	25.727273	7.566731	15.0	19.25	24.5	32.25	39.0	22.0	79.363636	10.504174	61.0	73.0	77.0	85.75	99.0
1	81.0	55.296296	8.988109	39.0	48.00	54.0	62.00	76.0	81.0	49.518519	6.530909	34.0	44.0	50.0	55.00	61.0
2	39.0	86.538462	16.312485	69.0	75.50	79.0	95.00	137.0	39.0	82.128205	9.364489	63.0	74.5	83.0	90.00	97.0
3	23.0	26.304348	7.893811	15.0	19.50	25.0	33.00	39.0	23.0	20.913043	13.017167	3.0	9.5	17.0	33.50	40.0
4	35.0	88.200000	16.399067	70.0	77.50	85.0	97.50	137.0	35.0	17.114286	9.952154	1.0	10.0	16.0	23.50	39.0

In [59]:

pd.pivot_table(df, index="Group", aggfunc = np.mean)

Out[59]:

	Age	Annual Income (k$)	CustomerID	Spending Score (1-100)
Group
0	25.272727	25.727273	23.090909	79.363636
1	42.716049	55.296296	86.320988	49.518519
2	32.692308	86.538462	162.000000	82.128205
3	45.217391	26.304348	23.000000	20.913043
4	41.114286	88.200000	164.371429	17.114286

In [60]:

pd.pivot_table(df, index="Group", aggfunc = np.mean, values = ['Annual Income (k$)','Spending Score (1-100)'])

Out[60]:

	Annual Income (k$)	Spending Score (1-100)
Group
0	25.727273	79.363636
1	55.296296	49.518519
2	86.538462	82.128205
3	26.304348	20.913043
4	88.200000	17.114286