What might be wrong with this query:
select count(customer_email) as num_prev
from _pj_cust_email_by_date
where order_date < '2011-02'
and customer_email is not null
group by customer_email having count(orde开发者_StackOverflow中文版r_date) > 0;
Which returns row results such as:
1
2
3
2
1
5
4
When I'm trying to get a full count of how many customers in total purchased during the specified date range?
_pj_cust_email_by_date
is a view that returns only email address and order date in YYYY-MM-DD format. I do not have access to use anything save for this view.
The GROUP BY
is causing that.
It causes one result row to be returned per group, in this for each distinct value of customer_email
.
If you want the total number of distinct email addresses, then you need to drop the GROUP BY
clause and change the COUNT
to COUNT(DISTINCT customer_email)
.
You need to subquery it further
select count(*) CustomerCount
from (
select count(customer_email) as num_prev
from _pj_cust_email_by_date
where order_date < '2011-02'
and customer_email is not null
group by customer_email having count(order_date) > 0;
) as innercount
That would normally be the approach, but since you're using having count(order_date) > 0
, I think you only need
select count(distinct customer_email) as num_prev
from _pj_cust_email_by_date
where order_date < '2011-02' and customer_email is not null
Because the HAVING clause will never detail with empty order_dates which makes the HAVING clause a dud, actually.
精彩评论