Reputation: 1846
I have the following table:
id partid orderdate qty price
1 10 01/01/2017 10 3
2 10 02/01/2017 5 9
3 11 01/01/2017 0.5 0.001
4 145 02/01/2017 5 18
5 10 12/12/2016 8 7
6 10 05/07/2010 81 7.5
Basically I want to compare the most recent purchasing of parts to the other purchasing of the same part in a period of 24 months
. For that matter compare id=2
to id = 1,5
.
I want to check if the price of the latest orderdate
(per part) is larger than the average price of that part in the last 24 months.
So first I need to calculate the avg price:
partid avgprice
10 (3+9+7)/3=6.33 (7.5 is out of range)
11 0.001
145 18
I also need to know the latest orderdate of each part:
id partid
2 10
3 11
4 145
and then I need to check if id=2, id=3, id=6 (latest purchases) are bigger than the average. If they are I need to return their partid
.
So I should have something like this:
id partid avgprice lastprice
2 10 6.33 9
3 11 0.001 0.001
4 145 18 18
Finally I need to return partid=10
since 9>6.33
Now to my questions...
I'm not sure how I can find the latest order in PostgreSQL
.
I tried:
select id, distinct partid,orderdate
from table
where orderdate> current_date - interval '24 months'
order by orderdate desc
This gives :
ERROR: syntax error at or near "distinct".
I'm a bit of a lost here. I know what I want to do but I cant translate it to SQL. Any one can help?
Upvotes: 1
Views: 2323
Reputation: 94884
Get the avarage per part and the last order per price and join these:
select
lastorder.id,
lastorder.partid,
lastorder.orderdate,
lastorder.price as lastprice,
avgorder.price as avgprice
from
(
select
partid,
avg(price) as price
from mytable
where orderdate >= current_date - interval '24 months'
group by partid
) avgorder
join
(
select distinct on (partid)
id,
partid,
orderdate,
price
from mytable
order by partid, orderdate desc
) lastorder on lastorder.partid = avgorder.partid
and lastorder.price > avgorder.price;
Upvotes: 2
Reputation: 1334
This can be solved without distinct
(which is heavy on the DB anyways):
with avg_price as (
select partid, avg(price) as price
from table
where orderdate> current_date - interval '24 months'
group by partid
)
select f.id, f.partid, av.price, f.price
from (
select id, partid, orderdate, price, rank() over (partition by partid order by orderdate desc)
from table
) as f
join avg_price av on f.partid = av.partid
where f.rank = 1
and av.price < f.price
Upvotes: 1