solve problem 39

Author: a97410985

MySol/Average selling price.md

@@ -0,0 +1,48 @@
+把資料庫建起來的SQL
+現在建table會在名稱後綴題目編號，這樣避免table名稱撞名
+```SQL
+create table Prices_39 (
+	product_id int,
+    start_date date,
+    end_date date,
+    price int,
+    PRIMARY KEY (product_id, start_date, end_date)
+);
+
+INSERT INTO Prices_39 (product_id, start_date, end_date, price) VALUES
+(1, "2019-02-17", "2019-02-28", 5),
+(1, "2019-03-01", "2019-03-22", 20),
+(2, "2019-02-01", "2019-02-20", 15),
+(2, "2019-02-21", "2019-03-31", 30);
+
+
+create table UnitsSold_39 (
+    product_id int,
+    purchase_date date,
+    units int
+);
+
+INSERT INTO UnitsSold_39 (product_id, purchase_date, units) VALUES
+(1, "2019-02-25", 100),
+(1, "2019-03-01", 15),
+(2, "2019-02-10", 200),
+(2, "2019-03-22", 30);
+```
+
+查詢的SQL
+```SQL
+SELECT product_id, round(SUM(temp.units * temp.price) / SUM(temp.units), 2) average_price FROM
+(SELECT unitssold_39.product_id, price, units FROM unitssold_39 
+LEFT JOIN prices_39 ON unitssold_39.product_id = prices_39.product_id 
+AND unitssold_39.purchase_date between prices_39.start_date AND prices_39.end_date) AS temp
+group by product_id;
+```
+
+先join`UnitsSold` table和`Prices` table，用left join可以確保`UnitSold`都會在，這題先把所有購買紀錄都對上對應時間區間的間隔，暫存為table temp(subquery's result)，在對temp table做query。
+如果是`natural join`就不用在合併條件那邊加上product_id要相同，而且會使用一般的where clause來過濾日期不相符的rows。如下
+```SQL
+SELECT product_id, round(SUM(temp.units * temp.price) / SUM(temp.units), 2) average_price FROM
+(SELECT unitssold_39.product_id, price, units FROM unitssold_39 
+natural join prices_39 where unitssold_39.purchase_date between prices_39.start_date AND prices_39.end_date) AS temp
+group by product_id;
+```