通过一个简单的电商零售数据集，了解数据分析流程

import pandas as pd
import matplotlib.pyplot as plt
import os
import numpy as np

import plotly as py
import plotly.graph_objs as go
py.offline.init_notebook_mode()
pyplot = py.offline.iplot

os.chdir(r'C:\Users\W\Desktop')

online_data = pd.read_csv('data.csv',encoding = 'ISO-8859-1',dtype = {'CustomerID':str})
online_data.head()

online_data.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 541909 entries, 0 to 541908
Data columns (total 8 columns):
InvoiceNo      541909 non-null object
StockCode      541909 non-null object
Description    540455 non-null object
Quantity       541909 non-null int64
InvoiceDate    541909 non-null object
UnitPrice      541909 non-null float64
CustomerID     406829 non-null object
Country        541909 non-null object
dtypes: float64(1), int64(1), object(6)
memory usage: 33.1+ MB

查看退货订单

online_data[online_data['InvoiceNo'].str[0]=='C']

2数据清洗

CustomerID 缺失率较高

# 数据缺失率
online_data.apply(lambda x:sum(x.isnull())/len(x),axis=0)

InvoiceNo      0.000000
StockCode      0.000000
Description    0.002683
Quantity       0.000000
InvoiceDate    0.000000
UnitPrice      0.000000
CustomerID     0.249267
Country        0.000000
dtype: float64

删除带有空值的行

# 防止修改df1对online_data造成修改，使用copy对数据做一份拷贝
df1 = online_data.dropna(how='any').copy()

df1.head()

# 将订单日期由字符串转换为标准日期格式
# errors = 'coerce' 错误转换为缺失值（容错处理）
df1['InvoiceDate'] = pd.to_datetime(df1['InvoiceDate'],errors = 'coerce')
# 提取日期部分
df1['InvoiceDate'] = df1['InvoiceDate'].dt.date
df1.head()

# 再次转换日期格式 object->datetime64[ns]
# df1['InvoiceDate'] = pd.to_datetime(df1['InvoiceDate'],errors = 'coerce')
df1.info()

<class 'pandas.core.frame.DataFrame'>
Int64Index: 406829 entries, 0 to 541908
Data columns (total 8 columns):
InvoiceNo      406829 non-null object
StockCode      406829 non-null object
Description    406829 non-null object
Quantity       406829 non-null int64
InvoiceDate    406829 non-null object
UnitPrice      406829 non-null float64
CustomerID     406829 non-null object
Country        406829 non-null object
dtypes: float64(1), int64(1), object(6)
memory usage: 27.9+ MB

销售金额

df1['Price'] = df1.apply(lambda x:x[3]*x[5],axis=1)
df1.head()

3数据分析和可视化

1.购买商品前十的国家是？

quantity_first_10 = df1[df1['Quantity']>0].groupby('Country').sum()['Quantity'].sort_values(ascending=False).head(10)
print(quantity_first_10)

Country
United Kingdom    4269472
Netherlands        200937
EIRE               140525
Germany            119263
France             111472
Australia           84209
Sweden              36083
Switzerland         30083
Spain               27951
Japan               26016
Name: Quantity, dtype: int64

trace_basic = [go.Bar(x = quantity_first_10.index.tolist(), 
                      y = quantity_first_10.values.tolist(),
                      marker = dict(color = 'red'),
                      opacity = .5)]
layout = go.Layout(title = '购买商品前十的国家',xaxis = dict(title='国家'))
figure_basic = go.Figure(data = trace_basic,layout = layout)
pyplot(figure_basic)

2.交易额前十的国家是？

price_first_10 = df1[df1['Quantity']>0].groupby('Country').sum()['Price'].sort_values(ascending=False).head(10)
print(quantity_first_10)

Country
United Kingdom    4269472
Netherlands        200937
EIRE               140525
Germany            119263
France             111472
Australia           84209
Sweden              36083
Switzerland         30083
Spain               27951
Japan               26016
Name: Quantity, dtype: int64

trace_basic = [go.Bar(x = price_first_10.index.tolist(), 
                      y = price_first_10.values.tolist(),
                      marker = dict(color = 'red'),
                      opacity = .5)]
layout = go.Layout(title = '交易额前十的国家',xaxis = dict(title='国家'))
figure_basic = go.Figure(data = trace_basic,layout = layout)
pyplot(figure_basic)

3.哪些月份销售较佳？

df1.head()

df1['month'] = pd.to_datetime(df1['InvoiceDate'],errors = 'coerce').dt.month

df1.head()

month_quantity = df1[df1['Quantity']>0].groupby('month').sum()['Quantity'].sort_values(ascending=False)
print(month_quantity)

month
11    681888
12    599693
10    593908
9     544899
8     398938
5     373685
7     369432
6     363699
1     349147
3     348544
4     292225
2     265638
Name: Quantity, dtype: int64

seaborn 绘图更简便

import seaborn as sns
sns.set(style = 'darkgrid',context='notebook',font_scale=1.2)

month_quantity.plot(kind='bar')
plt.xticks(rotation=45) #标签旋转45

4.客单价多少？

三种维度

sumPrice = df1[df1['Quantity']>0]['Price'].sum()
sumPrice

8911407.904

1.按每次订单

#订单有重复 未去重
sum_InvoiceNo = df1[df1['Quantity']>0]['InvoiceNo'].count()
sum_InvoiceNo

avgPrice = sumPrice/sum_InvoiceNo
avgPrice

22.394748504739596

2.按每个订单

# 去除重复订单号
sum_InvoiceNo = df1[df1['Quantity']>0]['InvoiceNo'].drop_duplicates().count()
sum_InvoiceNo

avgPrice = sumPrice/sum_InvoiceNo
avgPrice

480.7621873111782

3.按每个客户

# 客单价
sum_CustomerID = df1[df1['Quantity']>0]['CustomerID'].drop_duplicates().count()
sum_CustomerID

avgPrice = sumPrice/sum_CustomerID
avgPrice

2053.7930177460244

5.用户消费行为分析

customer = df1[df1['Quantity']>0].groupby('CustomerID').agg({'InvoiceNo':'nunique',
                                                           'Quantity':np.sum,
                                                           'Price':np.sum})
customer