1.数据清洗

步骤：

1.查询charge_record表业务类型为充值且订单状态为成功的数据

2.将上述数据转移到本地数据库

使用如下脚本：

# coding=utf-8
import pymysql

# 原数据库链接
db1 = pymysql.connect(
    host='***',
    port=3306, user='***',
    passwd='***',
    db='***',
    charset='utf8')
cursor1 = db1.cursor()
# 定义查询语句
len1 = cursor1.execute('select uid,update_time from charge_record where buss_type=0 and charge_status=2 and charge_prod_id is not null')

# 迁移数据库链接
db2 = pymysql.connect(
    host='127.0.0.1',
    port=3306, user='root',
    passwd='123456',
    db='test',
    charset='utf8')
cursor2 = db2.cursor()
# 批量插入语句
sql = 'insert into charge_record(uid,update_time) value(%s, %s)'

# 导入全部数据
data2 = cursor1.fetchall()
cursor2.executemany(sql, data2)

# 提交到数据库
db2.commit()

# 关闭数据库连接
db1.close()
db2.close()

2.计算留存率

使用的MySQL语句如下：

USE test;

-- 计算首单时间

SELECT uid,min(update_time) u_time FROM charge_record GROUP BY uid;

-- 重采首付时间

SELECT a.uid,b.u_time,TIMESTAMPDIFF(MONTH,b.u_time,a.update_time) m_diff,CONCAT(YEAR(b.u_time),"年",MONTH(b.u_time),"月") y_m FROM charge_record a

LEFT JOIN (

SELECT uid,min(update_time) u_time FROM charge_record GROUP BY uid LIMIT 0,7000

) b on a.uid=b.uid WHERE b.u_time is NOT NULL;

-- 计算留存量

CREATE table cohort as

SELECT c.y_m "首付月份",c.m_diff"月份差",COUNT(DISTINCT c.uid) "留存量" FROM (

SELECT a.uid,b.u_time,TIMESTAMPDIFF(MONTH,b.u_time,a.update_time) m_diff,CONCAT(YEAR(b.u_time),"年",MONTH(b.u_time),"月") y_m FROM charge_record a

LEFT JOIN (

SELECT uid,min(update_time) u_time FROM charge_record GROUP BY uid

) b on a.uid=b.uid WHERE b.u_time is NOT NULL ) c GROUP BY c.y_m,c.m_diff;

-- 计算留存率

SELECT c.`首付月份`,CONCAT(ROUND((c.`留存量`/m.`留存量`)*100,2),"%") 留存率 FROM cohort c

LEFT JOIN (

SELECT 首付月份,留存量 FROM cohort WHERE `月份差`=0

) m on c.`首付月份`=m.`首付月份`;

-- 留存率进阶版

SELECT

n.`首付月份`,

AVG(n.`留存量`) "本月新增",

CONCAT(sum(n.`+1月`),"%") "+1月",

CONCAT(sum(n.`+2月`),"%") "+2月",

CONCAT(sum(n.`+3月`),"%") "+3月",

CONCAT(sum(n.`+4月`),"%") "+4月",

CONCAT(sum(n.`+5月`),"%") "+5月"

FROM (
# 一级子查询：转置表格，将月份差作为列名
SELECT

a.`首付月份`,

a.`留存量`,

CASE a.`月份差` when 1 THEN a.`留存率` ELSE 0 END "+1月",

CASE a.`月份差` when 2 THEN a.`留存率` ELSE 0 END "+2月",

CASE a.`月份差` when 3 THEN a.`留存率` ELSE 0 END "+3月",

CASE a.`月份差` when 4 THEN a.`留存率` ELSE 0 END "+4月",

CASE a.`月份差` when 5 THEN a.`留存率` ELSE 0 END "+5月"

FROM(

# 二级子查询：计算留存率

SELECT a.`首付月份`,b.`留存量`,a.`月份差`,ROUND((a.`留存量`/b.`留存量`)*100,2) 留存率

FROM cohort a

LEFT JOIN (
# 三级子查询：查询首月用户量
SELECT `首付月份`,`留存量`

FROM cohort

WHERE cohort.`月份差`=0

) b
on a.`首付月份`=b.`首付月份`
) a
) n
GROUP BY n.`首付月份`;

最终结果如下：