Question

AVG与上一年相比记录增加

  SELECT TO_CHAR (Application_Date, 'yyyy'), 
         COUNT (*)
    FROM Services
GROUP BY TO_CHAR (Application_Date, 'yyyy')
ORDER BY 1 DESC;

AVG与上一年相比记录增加

示例：

2018 ----- 3000 records - 50% increase
2017 ----- 2000 records - 100% increase

2016 ----- 1000 records - 0% increase

Answer 1

假设参考平均值是一条红色的鲱鱼，而您的35％的示例是一个错字...

您要比较两行中的值，因此可以使用lag()函数通过以下命令查看上一年的数据：

LAG(COUNT(*)) OVER (ORDER BY TO_CHAR (Application_Date, 'yyyy'))

然后使用当年和上一年的计数正常计算百分比变化：

SELECT TO_CHAR (Application_Date, 'yyyy') AS year,
  COUNT(*) AS year_count,
  100 * (COUNT(*) - LAG(COUNT(*)) OVER (ORDER BY TO_CHAR (Application_Date, 'yyyy')))
    / LAG(COUNT(*)) OVER (ORDER BY TO_CHAR (Application_Date, 'yyyy')) AS percentage_change
FROM Services
GROUP BY TO_CHAR (Application_Date, 'yyyy')
ORDER BY 1 DESC;

YEAR YEAR_COUNT PERCENTAGE_CHANGE
---- ---------- -----------------
2018       5000                25
2017       4000                 0
2016       4000

您可以根据需要设置格式的计算值（例如，是否希望将％符号用作结果的一部分，例如将其设置为字符串，然后不清楚），并且可以使用合并替换第一年的空值如果需要，则为零。

您还可以使用CTE或内联视图仅进行一次滞后：

SELECT year,
  year_count,
  100 * (year_count - prev_year_count) / prev_year_count AS percentage_change
FROM (
  SELECT TO_CHAR (Application_Date, 'yyyy') AS year,
    COUNT(*) AS year_count,
    LAG(COUNT(*)) OVER (ORDER BY TO_CHAR (Application_Date, 'yyyy')) AS prev_year_count
  FROM Services
  GROUP BY TO_CHAR (Application_Date, 'yyyy')
)
ORDER BY 1 DESC;

个人使用年份值时，我倾向于将它们视为数字而不是字符串，因此我将使用提取：

select year,
  year_count,
  100 * (year_count - prev_year_count) / prev_year_count as percentage_change
from (
  select extract(year from application_date) as year,
    count(*) as year_count,
    lag(count(*)) over (order by extract(year from application_date)) as prev_year_count
  from services
  group by extract(year from application_date)
)
order by 1 desc;

      YEAR YEAR_COUNT PERCENTAGE_CHANGE
---------- ---------- -----------------
      2018       5000                25
      2017       4000                 0
      2016       4000

db<>fiddle

与上一年相比，AVG记录增加

1 个答案: