SQL / HIVE：如何计算购买天数

Question

SQL/Hive: I am looking to calculate # of days to purchase by a visitor. Here is how my data looks like

date    visitor orders
1-Jan   A   0  
1-Jan   B   0  
4-Jan   B   1  
5-Jan   A   0  
12-Jan  A   1

This is the result I am expecting:

Days to purchase    count of visitors
0   0
1   0 
2   0
3   1
4   0
5   0
.   .
.   .
.   .
11  1

Any help ?

Answer 1

if I understood you correctly: what you need to do is to find the minmum date for each combination of visitor+orders

select visitor,orders,min(date) as min.date from table group by visitor,orders

this should give something like:

visitor orders min.date
  A         0  1-Jan 
  B         0  1-Jan
  B         1  4-Jan
  A         1  12-Jan

this table (lets call it tbl) can be self joined to give

select A.visitor,datediff(day,purchase.date,first.visit) as days.to.purchase 
from (select visitor,min.date as first.visit from tbl where orders=0) A 
inner join (select visitor,min.date as purchase.date from tbl where orders=1) B
on A.visitor=B.visitor

now, wrap this query with an outer query to count the visitors with same datediffs:

 select days.to.purchase,count(visitors) as visitors from 
 (select A.visitor,datediff(day,purchase.date,first.visit) as days.to.purchase 
    from (select visitor,min.date as first.visit from tbl where orders=0) A 
    inner join (select visitor,min.date as purchase.date from tbl where orders=1) B
    on A.visitor=B.visitor
) joined
group by days.to.purchase order by days.to.purchase

hope I understood you correctly. I'm not sure this is the right solution but you didn't give me much to start with :)

the full solution could be:

 select days.to.purchase,count(visitors) as visitors from 
 (select A.visitor,datediff(day,purchase.date,first.visit) as days.to.purchase 
    from 
(select visitor,min.date as first.visit from 
(select visitor,orders,min(date) as min.date from table group by visitor,orders) tbl where orders=0) A 
    inner join 
(select visitor,min.date as purchase.date from 
(select visitor,orders,min(date) as min.date from table group by visitor,orders) tbl where orders=1) B
    on A.visitor=B.visitor
) joined
group by days.to.purchase order by days.to.purchase