穀粒影音：業務處理

統計視頻觀看數Top10

• 思路：使用order by按照 views 字段做一個全局排序，設置只顯示前10條即可
  實現SQL：

select videoid,uploader,age,category,length,views,rate,ratings,comments
from guliyingyin_video_orc
order by views desc
limit 10;

![在這裏插入圖片描述](https://img-blog.csdnimg.cn/20200222225830179.png?x-oss-process=image/watermark,type_ZmFuZ3poZW5naGVpdGk,shadow_10,text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L2xpYW5naGVjYWk1MjE3MTMxNA==,size_16,color_FFFFFF,t_70)

• 結果：
  

統計視頻類別熱度Top10

思路：炸開數組”視頻類別“字段，然後按照類別分組，最後按照熱度(視頻個數)排序。

因爲當前表結構爲：一個視頻對應一個或多個類別。所以如果要 group by 類別，需要先將類別進行列轉行(展開)，然後再進行count即可

• 分析過程：

1. 炸開類別，記爲臨時表t1

   select videoId,catName
   from guliyingyin_video_orc
   lateral view explode(category) tb_category as catName; — t1

2. 按照類別 group by 聚合，然後count組內的videoId個數，記爲臨時表t2

   select catName,count(*) hot
   from t1
   group by catName; —t2

3. 最後按照熱度排序，顯示前10條。

   select catName, hot
   from t2
   order by hot desc
   limit 10;

• 完整的SQL：

  SELECT
  	catName,
  	hot 
  FROM
  	(
  	SELECT
  		catName,
  		count( * ) hot 
  	FROM
  		( SELECT videoId, catName FROM guliyingyin_video_orc lateral VIEW explode ( category ) tb_category AS catName ) t1 
  	GROUP BY
  		catName 
  	) t2 
  ORDER BY
  	hot DESC 
  	LIMIT 10;
  

將完整的SQL語句保存到guliyingying.sql文件中，並上傳到Linux。然後執行命令：

結果：

統計視頻觀看數Top20所屬類別

• 思路：

1. 先找到觀看數最高的20個視頻，記爲臨時表t1

   select videoid,views,category
   from guliyingyin_video_orc
   order by views desc
   limit 20; ----t1

2. 把這20條信息中的category分裂出來(列轉行)，記爲臨時表t2

   select videoid,catName
   from t1
   lateral view explode(category) tb_category as catName; — t2

3. 去重

   select distinct catName
   from t2;

• 完整SQL

SELECT DISTINCT
	catName 
FROM
	(
	SELECT
		videoid,
		catName 
	FROM
	( SELECT videoid, views, category FROM guliyingyin_video_orc ORDER BY views DESC LIMIT 20 ) t1 lateral VIEW explode ( category ) tb2_category AS catName 
	) t2;

將完整的SQL語句保存到guliyingying.sql文件中，並上傳到Linux。然後執行命令：

結果：

統計視頻觀看數Top50所關聯視頻的類別的Rank

思路分析：

思路：

1. 查詢出觀看數最多的前50個視頻的所有信息(包含每個視頻對應的關聯視頻)，記爲臨時表t1

   select videoId, views, category, relatedId
   from guliyingyin_video_orc
   order by views desc
   limit 50; ----t1

2. 炸裂關聯視頻id：將找到的50條視頻信息的相關視頻的relatedId列轉行，記爲臨時表t2

   select distinct videoId_name
   from t1
   lateral view explode(relatedId) tb_relatedId as videoId_name; ----t2

3. 將關聯視頻的id和guliyingyin_video_orc表進行inner join操作，得到每個關聯視頻id的詳細數據，記爲臨時表t4

   select *
   from t2
   inner join guliyingyin_video_orc t3
   on t2.videoId_name=t3.videoId; ---- t4

4. 炸裂關聯視頻的類別

   select *
   from t4
   lateral view explode(category) tb_category as catName; ----t5

5. 統計類別個數

   select catName, count(*) hot
   from t5
   group by catName; ----t6

6. 統計類別的熱度排名

   select *
   from t6
   order by hot desc;

• 完整SQL：

SELECT
	* 
FROM
	(
	SELECT
		catName,
		count( * ) hot 
	FROM
		(
		SELECT
			* 
		FROM
			(
			SELECT
				* 
			FROM
				(
				SELECT DISTINCT
					videoId_name 
				FROM
					( SELECT videoId, views, category, relatedId FROM guliyingyin_video_orc ORDER BY views DESC LIMIT 50 ) t1 lateral VIEW explode ( relatedId ) tb_relatedId AS videoId_name 
				) t2
				INNER JOIN guliyingyin_video_orc t3 ON t2.videoId_name = t3.videoId 
			) t4 lateral VIEW explode ( category ) tb_category AS catName 
		) t5 
	GROUP BY
		catName 
	) t6 
ORDER BY
	hot DESC;

將完整的SQL語句保存到guliyingying.sql文件中，並上傳到Linux。然後執行命令：

結果：

---

統計每個類別中的視頻熱度Top10，以Music爲例

思路：要想統計Music類別中的視頻熱度Top10，需要先找到Music類別，那麼就需要將category炸開。

1. 創建一個臨時表用於存放categoryId炸開的數據

   create table guliyingyin_category_orc(
   videoId string, 
   uploader string, 
   age int, 
   categoryId string, 
   length int, 
   views int, 
   rate float, 
   ratings int, 
   comments int,
   relatedId array<string>
   )
   row format delimited 
   fields terminated by "\t"
   collection items terminated by "&"
   stored as orc;
   

2. 向category展開的表中插入數據。

   insert overwrite table guliyingyin_category_orc
   select  videoid,uploader,age,categoryId,length,views,rate,ratings,comments,relatedId
   from guliyingyin_video_orc
   lateral view explode(category) tb_category as categoryId;
   

3. 統計對應類別（Music）中的視頻熱度。

   select  videoId,views
   from guliyingyin_category_orc
   where categoryId = "Music"
   order by views desc
   limit 10;
   

   
   結果：
   

統計每個類別中視頻流量Top10，以Music爲例

select  videoId,ratings
from guliyingyin_category_orc
where categoryId = "Music"
order by ratings desc
limit 10;

結果：

統計上傳視頻最多的用戶Top10以及他們上傳的觀看次數在前20的視頻

• 分析

1. 上傳視頻最多的用戶Top10，記爲表t1

   select uploader
   from guliyingyin_user_orc
   order by videos desc
   limit 10; -----t1
   

2. 觀看次數在前20的視頻

   select t1.uploader,t2.videoId,views
   from t1 inner join guliyingyin_video_orc t2
   on t1.uploader= t2.uploader 
   order by views desc
   limit 20;
   

• 完整SQL語句

SELECT
	t1.uploader,
	t2.videoId,
	views 
FROM
	( SELECT uploader FROM guliyingyin_user_orc ORDER BY videos DESC LIMIT 10 ) t1
	INNER JOIN guliyingyin_video_orc t2 ON t1.uploader = t2.uploader 
ORDER BY
	views DESC 
	LIMIT 20;

• 結果
  

統計每個類別視頻觀看數Top10

• 思路：

1. 先得到categoryId展開的中間表

2. 子查詢按照categoryId進行分區，然後分區內排序，並生成遞增數字，該遞增數字這一列起名爲rank列

   select videoid,categoryid,views,
   row_number() over(partition by categoryid order by views desc) rank234
   from guliyingyin_category_orc;
   

3. 通過子查詢產生的臨時表，查詢rank值小於等於10的數據行即可

   select * 
   from t1
   where rank <= 10;
   

• 完全SQL語句：

  SELECT * 
  FROM
  	( SELECT videoid, categoryid, views, row_number ( ) over ( PARTITION BY categoryid ORDER BY views DESC ) rank FROM guliyingyin_category_orc ) t1 
  WHERE
  	rank234 <= 10;
  

• 結果