Electronic Product Sales Data Analysis

@[TOC] Electronic product sales data analysis catalog

Introduction to Data Parameters

Import of database

Handling of duplicate values ​​with missing values

Calculate the total amount of monthly consumption

Total monthly consumption

user analysis

brand sales

Visualization of user consumption behavior

Data parameter introduction

event_time -购买时间
order_id -订单编号
product_id -产品编号
category_id -产品的类别ID
category_code -产品的类别分类法（代码名称）
brand -品牌名称
price -产品价格
user_id -用户ID

Import of database

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import datetime as datetime
import math
plt.rcParams['font.sans-serif'] = ['SimHei']  #设置中文字体为黑体
plt.rcParams['axes.unicode_minus'] = False #正常显示负号
pd.set_option('display.float_format',lambda x : '%.2f' % x)#pandas禁用科学计数法

read data

data = pd.read_csv('F:/数据集/kz.csv')
data.head()
data.info()

data preprocessing

data.isnull().sum()
# 删除所有有缺失值的行
data = data.dropna()
# 将时间列更改为时间类型
data['event_time'] = pd.to_datetime(data['event_time'])

Analyze user spending trends by month

# 去掉category_id、user_id的小数
data['category_id'] = data['category_id'].astype(np.int64,errors='ignore')
# data.info()
data['category_id'].dtype

Calculate the total amount of monthly consumption

# 解释：消费金额以用户Id来分组并求和
# 先计算出每个id同一产品的消费次数,增加购买数量和购买总价的列
# 购买数量的列
buy_cnt = data.groupby(['order_id','product_id']).agg(buy_cnt=('user_id','count'))
data = pd.merge(data,buy_cnt,on=['order_id','product_id'],how='inner')
data = data.drop_duplicates().reset_index(drop=True)
# 购买总价列
data['buy_amount']=data['buy_cnt']*data['price']
# 每月消费总金额
# 计算Gmv是商品交易总额
data['day'] = data['event_time'].apply(lambda x:x.day)
# 计算每月的GMV
GMV_month = data.groupby(data['month']).agg(GMV=('buy_amount','sum'))
GMV_month.sort_values(by='GMV',ascending=False)
# 8月是商品交易总额最大的时候，从6月开始增加，从8月开始减小

Total Sales Visualization

plt.plot(GMV_month.index,GMV_month['GMV'])
plt.title('每月销售总额')
plt.show()

The total sales reached the highest value in August, and the sales rose rapidly from June to August

user analysis

Number of consumers per month

data['user_id'].sort_values().unique()
# User consumption times
num_cnt = data.groupby(['user_id','month'])['price'].count() num_cnt # Monthly active number month_num_count = data.groupby(['month'])['price'].agg([
'
sum
','count']).reset_ind ex().rename(columns={'sum':'sales','month':'month','count':'active number'})
month_num_count

Visualization of user consumption

plt.plot(month_num_count.index,month_num_count['活跃人数'])
plt.title('每月消费人数')
plt.xlabel('消费月份')
plt.ylabel('活跃人数')
plt.legend('人数')

It began to rise in March, and after a decline in April, it began to rise rapidly until July. From July
to October is the peak consumption period, and the consumption and active people in other months are relatively small

Brand sales analysis

groud_sum = data.groupby('brand')['price'].agg('sum').reset_index().rename(columns={'brand':'品牌','price':'销售额'})
groud_sum20 = groud_sum.sort_values(by='销售额',ascending=False)[:20]
groud_sum20
# 列举出前二十名
sns.barplot(x='销售额',y='品牌',data=groud_sum20)

Per capita sales visualization

brand_cnt['sum_mean'] = groud_sum['销售额']/brand_cnt['用户数量']
sns.barplot(x='sum_mean',y='品牌',data=brand_cnt.sort_values('sum_mean',ascending=False)[:20])

Visual analysis of user consumption behavior

num_cnt_sum= data.groupby('user_id'['price','order_id'].agg({'price':'sum','order_id':'count'}).rename(columns={'price':'消费金额','order_id':'消费次数'})
num_cnt_sum
# 绘制消费次数和消费金额之间的关系
sns.scatterplot(x='消费金额',y='消费次数',data=num_cnt_sum,color='g')
plt.title('消费次数和消费金额之间的关系')
sns.displot(data=num_cnt_sum.query('消费金额<10000'),x='消费金额')

User Consumption Behavior

# 查看新增人数
# 查看最小的消费时间
data.loc[data.year ==2020].groupby('user_id')['event_time'].min().value_counts().plot()
#顾客最后一次购买时间
last_buy = data.loc[data.year==2020].groupby('user_id')['event_time'].max().reset_index()
last_buy
# 第一次购买
first_mon_buy = data.groupby('user_id')['event_time'].min().apply(lambda x:x.month).value_counts().sort_values(ascending=False)
first_mon_buy.plot(kind='bar')
first_mon_buy.to_frame().reset_index().rename(columns={'index':'月份','event_time':'第一次购买人数'})

#We can see that there is not much difference between the time of the first purchase and the last purchase, so it can be determined that the user basically bought once

Statistics of new and old customers

# 统计新老顾客
new_old = data.groupby('user_id')['event_time'].agg(['min','max'])
new_old
new_old['is_new']=(new_old['min']==new_old['max'])
new_old.head()
#最大值和最小值一样的，是新客户，反之是老客户

Create a user data pivot table to manage users hierarchically

# 创建透视表
rfm =data.pivot_table(index='user_id',
                values={'event_time','price','order_id'},
                aggfunc={
                    'event_time':'max',
                    'price':'sum',
                    'order_id':'count'
                }).reset_index().rename(columns={'order_id':'购买次数','event_time':'最后一次购买时间','price':'购买总金额'})
rfm

Determining user types by FRM method

def rfm_func(x):
    level = x.apply(lambda x: '1' if x>=0 else '0')
    label = level['最后一次购买间隔']+level['购买次数']+level['购买总金额']
    d = {
        '111' : '重要价值客户'
        ,'011': '重要保持客户' 
        ,'101': '重要发展客户'
        ,'001': '重要挽留客户'
        ,'110': '一般价值客户'
        ,'010': '一般保持客户'
        ,'100': '一般发展客户'
        ,'000': '一般挽留客户'
    }
    result = d[label]
    return result
rfm['label']=rfm[['最后一次购买间隔','购买次数','购买总金额']].apply(lambda x:x -x.mean()).apply(rfm_func,axis=1)
rfm.head()

User Layered Visualization

It can be seen that general development customers and general retention customers account for the majority, while important customers actually only account for a small part