Selecting cities that have a combination of products in SQL

I have the following dataset, that have the prices for many products and cities:

CREATE TABLE my_table (
    the_id varchar(5) NOT NULL, 
    the_date timestamp NOT NULL,
    the_city varchar(5) NOT NULL,
    the_product varchar(1) NOT NULL,
    the_price numeric(4,2) NOT NULL 
);

INSERT INTO my_table
VALUES ('VIS01', '2019-05-02 09:00:00','LISBO','A',8), 
       ('VIS02', '2019-05-04 12:00:00','EVORA','A',7), 
       ('VIS03', '2019-05-05 18:00:00','LISBO','B',5), 
       ('VIS04', '2019-05-06 18:30:00','PORTO','B',5), 
       ('VIS05', '2019-05-15 12:05:00','PORTO','A',6), 
       ('VIS06', '2019-05-18 18:06:00','EVORA','B',4),
       ('VIS07', '2019-05-23 18:07:00','PORTO','C',10),
       ('VIS08', '2019-05-26 18:08:00','EVORA','E',14),
       ('VIS09', '2019-06-07 18:09:00','LISBO','B',5),
       ('VIS10', '2019-06-09 18:10:00','LISBO','D',11),
       ('VIS11', '2019-06-12 18:11:00','EVORA','C',10),
       ('VIS12', '2019-06-15 18:12:00','LISBO','D',11),
       ('VIS13', '2019-06-15 18:13:00','EVORA','A',7),
       ('VIS14', '2019-06-18 18:14:00','PORTO','A',6),
       ('VIS15', '2019-06-23 18:15:00','LISBO','A',8),
       ('VIS16', '2019-06-25 18:16:00','LISBO','A',8),
       ('VIS17', '2019-06-27 18:17:00','PORTO','B',5),
       ('VIS18', '2019-06-27 18:18:00','LISBO','D',11),
       ('VIS19', '2019-06-28 18:19:00','LISBO','A',8),
       ('VIS20', '2019-06-30 18:20:00','EVORA','B',4),
       ('VIS21', '2019-07-01 18:21:00','EVORA','A',6),
       ('VIS22', '2019-07-04 18:30:00','EVORA','D',10),
       ('VIS23', '2019-07-04 18:31:00','EVORA','B',3),
       ('VIS24', '2019-07-06 18:40:00','EVORA','C',11),
       ('VIS25', '2019-07-12 18:50:00','EVORA','E',14),
       ('VIS26', '2019-07-15 18:00:00','PORTO','A',7),
       ('VIS27', '2019-07-18 18:00:00','PORTO','C',11),
       ('VIS28', '2019-07-25 18:00:00','PORTO','B',4),
       ('VIS29', '2019-07-30 18:00:00','PORTO','A',9),
       ('VIS30', '2019-07-30 18:00:00','LISBO','A',7);

What I want is to have the average price for each month and each city but it has to follow a rule, that it should have the combination of certain products. Let’s say that the product should be A and B strictly, then the expected table is:

month     city     product    avg_price
2019-05   EVORA    A          7
2019-05   EVORA    B          4
2019-05   LISBO    A          8
2019-05   LISBO    B          5
2019-05   PORTO    A          6
2019-05   PORTO    B          5
2019-06   EVORA    A          7
2019-06   EVORA    B          4
2019-06   LISBO    A          8
2019-06   LISBO    B          5
2019-06   PORTO    A          6
2019-06   PORTO    B          5
2019-07   EVORA    A          6
2019-07   EVORA    B          3
2019-07   PORTO    A          8
2019-07   PORTO    B          4

As you see, in this case LISBO doesn’t appear in 2019-07 because the condition “A and B” strictly doesn’t hold.

I have tried this code:

select date_trunc('month', the_date) as month, the_city, the_product, AVG(the_price) avg_price
FROM my_table
where the_product in ('A','B')
group by month, the_city, the_product
order by month desc

But I don’t know how to apply the condition. Please, could you help me with this question? Thank you in advance.

Answer

if you need in rows then you need to group by product as well:

select
    the_city,
    date_trunc('month', the_date) as yearmonth,
    the_product ,
    avg(the_price) as a_avg
from
(
    select * 
    , sum(case when the_product = 'A' then 1 end) over (partition by date_trunc('month', the_date), the_city) acount
    , sum(case when the_product = 'B' then 1 end) over (partition by date_trunc('month', the_date), the_city) bcount
    from my_table
    where  the_product in ('A','B')
) t 
where t.acount > 0 and t.bcount > 0 
group by the_city, yearmonth,the_product
order by yearmonth, the_city;