淘宝用户行为数据分析

import pandas as pd;
data = pd.read_csv(r‘C:\Users\ASUS\Desktop\UserBehavior\user.csv‘)
data = data.sample(n=400000,replace=True,axis=0)
data.to_csv(r"C:\Users\ASUS\Desktop\UserBehavior\user1.csv")
data.head()
data.info()

ALTER TABLE userbehavior ADD Date VARCHAR(255);
ALTER TABLE userbehavior ADD TIME VARCHAR(255);

UPDATE userbehavior SET Date = from_unixtime(Timestamps,‘%y-%m-%d‘);
UPDATE userbehavior SET TIME = from_unixtime(Timestamps,‘%H:%I:%S‘);

DELETE FROM userbehavior WHERE date not BETWEEN ‘17-11-25‘ AND ‘17-12-03‘

SELECT Behavior_type, COUNT(*) FROM userbehavior GROUP BY Behavior_type ORDER BY Behavior_type DESC;

SELECT Behavior_type, COUNT(DISTINCT user_id) as COUNT FROM userbehavior GROUP BY Behavior_type ORDER BY COUNT desc; 

CREATE VIEW  用户流失 as select user_id,sum(case when behavior_type=‘pv‘ then 1 else 0 end) as PV,
              sum(case when behavior_type=‘fav‘ then 1 else 0 end) as fav,
              sum(case when behavior_type=‘cart‘ then 1 else 0 end) as cart,
              sum(case when behavior_type=‘buy‘ then 1 else 0 end) as buy
from userbehavior
group by user_id;

SELECT COUNT(user_id) FROM `用户流失` WHERE PV = 1

SELECT COUNT(User_ID) FROM `用户流失` WHERE PV = 0

SELECT COUNT(User_ID) FROM `用户流失` WHERE PV = 0 AND buy > 0

select count(user_ID)as 用户总数,
       sum(case when pv<>0 and fav=0 and cart=0 and buy=0 then 1 else 0 end)as 浏览后流失,
       sum(case when fav>0 and cart=0 and buy=0 then 1 else 0 end)as 收藏后流失,
       sum(case when fav=0 and cart>0 and buy=0 then 1 else 0 end)as 加购后流失,
       sum(case when fav>0 and cart>0 and buy=0 then 1 else 0 end)as 加购收藏后流失,
       sum(case when buy<>0 then 1 else 0 end)as 购买人数
from 用户流失;

SELECT HOUR(time) as 小时, 
sum(case WHEN Behavior_type = ‘pv‘ then 1 else 0 end) as 点击数,
sum(case WHEN Behavior_type = ‘fav‘ then 1 else 0 end) as 收藏数,
sum(case WHEN Behavior_type = ‘cart‘ then 1 else 0 end) as 加购数,
sum(case WHEN Behavior_type = ‘buy‘ then 1 else 0 end) as 购买数
FROM userbehavior GROUP BY 小时 ORDER BY 小时;

SELECT DAYOFWEEK(date) as 周几, 
sum(case WHEN Behavior_type = ‘pv‘ then 1 else 0 end) as 点击数,
sum(case WHEN Behavior_type = ‘fav‘ then 1 else 0 end) as 收藏数,
sum(case WHEN Behavior_type = ‘cart‘ then 1 else 0 end) as 加购数,
sum(case WHEN Behavior_type = ‘buy‘ then 1 else 0 end) as 购买数
FROM userbehavior GROUP BY 周几 ORDER BY 周几

-- 设置参数提取有购买行为的用户总数
set @userBuyNum = (
select count(distinct user_id) as 购买用户数
from userbehavior 
where Behavior_type = ‘buy‘);
-- 下面提取R+F分数
select r.*,f.frequency,f.freq_rank,
(
-- 下面是根据R的排名进行打分
(case when 
r.recent_rank <= @userBuyNum*1/4 then 4
when (r.recent_rank > @userBuyNum*1/4) and (r.recent_rank <= @userBuyNum*2/4) then 3
when (r.recent_rank > @userBuyNum*2/4) and (r.recent_rank <= @userBuyNum*3/4) then 2
else 1 end)+
-- 下面是根据F的排名进行打分
(case when 
f.freq_rank <= @userBuyNum*1/4 then 4
when (f.freq_rank > @userBuyNum*1/4) and (f.freq_rank <= @userBuyNum*2/4) then 3
when (f.freq_rank > @userBuyNum*2/4) and (f.freq_rank <= @userBuyNum*3/4) then 2
else 1 end)
) as user_value
from
-- 提取R的相关数据
(select a.*,(@rank:=@rank+1) as recent_rank
FROM
(SELECT user_id,datediff(‘2017-12-04‘,max(date)) as recent
from userbehavior 
where Behavior_type = ‘buy‘
group by user_id 
order by recent) as a,
(select @rank:=0) as b) as r
left JOIN
-- 提取F的相关数据
(select a.*,(@rank2:=@rank2+1) as freq_rank
FROM
(select user_id,count(Behavior_type) as frequency
from userbehavior    
where Behavior_type = ‘buy‘
group by user_id 
order by frequency DESC) as a,
(select @rank2:=0) as b) as f
on r.user_id = f.user_id
ORDER BY user_value DESC;