-
Notifications
You must be signed in to change notification settings - Fork 0
/
Clustering.py
53 lines (34 loc) · 1.47 KB
/
Clustering.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
from sklearn.preprocessing import StandardScaler
from sklearn.cluster import KMeans
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
df = pd.read_csv("../Week3/data/Online Retail Germany.csv")
df.head(5)
# converting the Invoice date column to datetime format
df['InvoiceDate'] = pd.to_datetime(df['InvoiceDate'])
# creating a new column `Recency` to calculate how recent a customer purchased
# from our shop
df['Recency'] = pd.Timestamp.now().normalize() - df['InvoiceDate']
#converting to number to normalize using Z-Score
df['Recency'].astype('int')
# only extracting number of days as an integer
df['Recency'] = df['Recency'].dt.days
# grouping data by customerId and aggregating over the total purchase, quantity, and how recent
# they made a purchase from the store
groupedData = df.groupby('CustomerID').agg({'UnitPrice' : 'sum',
'Quantity': 'sum',
'Recency' : 'min'})
# normalizing the data
groupedData[['UnitPrice', 'Quantity', 'Recency']] = StandardScaler().fit_transform(groupedData[['UnitPrice', 'Quantity', 'Recency']])
# initialising the KMeans clustering class
kmeans = KMeans(init = "random",
n_clusters = 3,
n_init = 10,
max_iter = 99,
random_state = 42)
kmeans.fit(groupedData)
# plotting the clusters
groupedData['cluster'] = kmeans.labels_
plt.scatter('Quantity', 'Recency', c = 'cluster', data=groupedData)
plt.show()