避免在嵌套SQL查询中计算空值

我有一个PostgreSQL计算 roi 其中的几个值 select 100 * ((SUM(time_product)/SUM(rt_id_count))*SUM(rental_cost))/SUM(sale_count) 使用case语句作为嵌套查询来定义和计算。但是,如果其中任何值为空 rental_cost 它计算 投资回报率 作为 zero ,而这个零值将被卷进平均值中,并向下倾斜平均值。在总计算中,如何将这些零视为空?

(我取出了一些嵌套查询,并用 [calculation] 在这个问题上;主要是因为它是一个很长的查询,我想过滤掉噪音,使整个结构更可见)

  (select 100 * ((SUM(time_product)/SUM(rt_id_count))*SUM(rental_cost))/SUM(sale_count)  FROM

  (select p2.id, p2.inventory_type, p2.working_value, COUNT(distinct p2.id),

  ----nested values calculated here
   (case p2.inventory_type 
      when 'set' then 
      [calculation A]
    else [calculation B] end) rental_cost,
    (case p2.inventory_type 
      when 'set' then 
      [caluculation A]
    else [calculation B]  end)   sale_count,
    (case p2.inventory_type 
      when 'set' then count(distinct rt.id)
      else 1 end 
      ) rt_id_count,
    (case p2.inventory_type 
      when 'set' then 
      [calculation ]  end)   time_total,
 [calculation]  time_product

  FROM warehouses w
    LEFT JOIN rfid_tags rt ON w.id = rt.location_id AND rt.location_type = 'Warehouse'
    LEFT JOIN products p2 ON rt.ancestor_product_id = p2.id 
    LEFT JOIN category_assignments ca  ON ca.product_id = p2.id
    LEFT JOIN categories c ON ca.category_id = c.id
    LEFT JOIN product_selections ps ON ps.rfid_tag_id = rt.id 

    WHERE 
      c.id=categories.id AND ca.primary = true  AND w.id=warehouses.id
      AND (select count(ps.id) from product_selections ps where ps.rfid_tag_id=rt.id)>0
      and p2.working_value>0 
    AND rt.location_id=w.id
    group by p2.id, p2.inventory_type, p2.working_value, c.sale_price_percentage, c.rental_price_percentage) Z) roi