如何选择不同日期的交换值？

关闭。这个问题需要更加突出重点。它目前不接受答案。
**想改进这个问题吗？**通过编辑这篇文章更新这个问题，使它只关注一个问题。

去年关门了。
改进这个问题
我使用的是clickhouse，我想知道这个“搜索词”在不同的日子里是不同的路径值
我的数据

date    searchword  path
1   2019-10-02  word1   url-1
2   2019-10-03  word1   url-2
3   2019-10-04  word1   url-1
4   2019-10-11  word1   url-1
5   2019-10-07  word2   url-4
6   2019-10-08  word3   url-5
7   2019-10-09  word3   url-5
8   2019-10-10  word3   url-5
9   2019-10-08  word4   url-5
10  2019-10-09  word4   url-5
11  2019-10-10  word4   url-5
12  2019-10-10  word4   url-6
13  2019-10-11  word4   url-5
14  2019-10-10  word5   url-7
15  2019-10-09  word5   url-8
16  2019-10-11  word5   url-7

期望结果

word1 
word5

但不是word4，因为没有值的交换，而是同一天的两个值
word1（两条路径，每天一条，第2行的值为url2，其他天，第1行和第3行的值为url1）
word5（多个值，第14行的值为url-7，第15行的值为url-8）。
我想我应该使用sequencematch或windowfunnel函数，知道吗？

尝试此查询：

SELECT searchword
FROM (
  /* intermediate result of SELECT below
  ┌─searchword─┬─date_paths_array──────────────────────────────────────────────────────────────────────────────────────────────┬─isExchanged─┐
  │ word3      │ [('2019-10-09',['url-5']),('2019-10-10',['url-5']),('2019-10-08',['url-5'])]                                  │           0 │
  │ word5      │ [('2019-10-10',['url-7']),('2019-10-11',['url-7']),('2019-10-09',['url-8'])]                                  │           1 │
  │ word1      │ [('2019-10-04',['url-1']),('2019-10-02',['url-1']),('2019-10-03',['url-2']),('2019-10-11',['url-1'])]         │           1 │
  │ word4      │ [('2019-10-08',['url-5']),('2019-10-10',['url-5','url-6']),('2019-10-09',['url-5']),('2019-10-11',['url-5'])] │           0 │
  └────────────┴───────────────────────────────────────────────────────────────────────────────────────────────────────────────┴─────────────┘
  */
  SELECT 
    searchword, 
    groupArray(date_paths) date_paths_array, 
    arrayFirstIndex((x, index) -> 
        index != length(date_paths_array) 
        AND hasAny(x.2, date_paths_array[index + 1].2) = 0, 
      date_paths_array, 
      arrayEnumerate(date_paths_array)) > 0 isExchanged
  FROM (    
    /* intermediate result of SELECT below
    ┌─searchword─┬─date_paths───────────────────────┐
    │ word1      │ ('2019-10-04',['url-1'])         │
    │ word3      │ ('2019-10-09',['url-5'])         │
    │ word5      │ ('2019-10-10',['url-7'])         │
    │ word1      │ ('2019-10-02',['url-1'])         │
    │ word4      │ ('2019-10-08',['url-5'])         │
    │ word2      │ ('2019-10-07',['url-4'])         │
    │ word5      │ ('2019-10-11',['url-7'])         │
    │ word5      │ ('2019-10-09',['url-8'])         │
    │ word3      │ ('2019-10-10',['url-5'])         │
    │ word1      │ ('2019-10-03',['url-2'])         │
    │ word3      │ ('2019-10-08',['url-5'])         │
    │ word1      │ ('2019-10-11',['url-1'])         │
    │ word4      │ ('2019-10-10',['url-5','url-6']) │
    │ word4      │ ('2019-10-09',['url-5'])         │
    │ word4      │ ('2019-10-11',['url-5'])         │
    └────────────┴──────────────────────────────────┘
    */
    SELECT searchword, (date, groupArray(path)) date_paths
    FROM (
      /* test dataset */    
      SELECT d.1 order_num, toDate(d.2) date, d.3 searchword, d.4 path
      FROM (  
        SELECT arrayJoin([
          (1, '2019-10-02', 'word1', 'url-1'),
          (2, '2019-10-03', 'word1', 'url-2'),
          (3, '2019-10-04', 'word1', 'url-1'),
          (4, '2019-10-11', 'word1', 'url-1'),
          (5, '2019-10-07', 'word2', 'url-4'),
          (6, '2019-10-08', 'word3', 'url-5'),
          (7, '2019-10-09', 'word3', 'url-5'),
          (8, '2019-10-10', 'word3', 'url-5'),
          (9, '2019-10-08', 'word4', 'url-5'),
          (10, '2019-10-09', 'word4', 'url-5'),
          (11, '2019-10-10', 'word4', 'url-5'),
          (12, '2019-10-10', 'word4', 'url-6'),
          (13, '2019-10-11', 'word4', 'url-5'),
          (14, '2019-10-10', 'word5', 'url-7'),
          (15, '2019-10-09', 'word5', 'url-8'),
          (16, '2019-10-11', 'word5', 'url-7')]) d))
    WHERE date >= '2019-10-01' AND date < '2019-11-01'      
    GROUP BY searchword, date)
  GROUP BY searchword  
  HAVING count() > 1)
WHERE isExchanged;
/* Result:
┌─searchword─┐
│ word5      │
│ word1      │
└────────────┘

* /

如何选择不同日期的交换值？

1条答案

相关问题

热门标签

最新问答