plsql - 如何开发统计模型来识别 peoplesoft 系统中的问题过程，该模型应该赋予最近运行比旧运行更多的权重

Question

下面的查询仅给出加权平均值和正常平均值之间的微小差异，什么应该是好的模型以及如何在 peoplesoft 中获得工作失败的概率

 SELECT DISTINCT TRUNC(SUM(ntile1) over(partition BY prcsname,status order by status)/SUM(ntile1) over(partition BY prcsname),2)*100 as weighted_pecentage,
 TRUNC(count(1) over(partition BY prcsname,status order by status)/count(1) over(partition BY prcsname),2)*100 as normal_percentage,
 prcsname,
 status
 FROM
 (SELECT 100-ntile(100) over (partition by prcsname order by prcsname ,prcsinstance DESC)AS ntile1,
 DECODE(runstatus,9,'Success','UnSuccess') AS status ,
 p.*
 FROM psprcsrqst p
 WHERE prcsname LIKE '%'
 ORDER BY
 prcsinstance desc, prcsname,status DESC
 ) p2 order by prcsname,
 status;
 
 !! detail
 SELECT 100-ntile(100) over (partition by prcsname order by prcsinstance desc )AS ntile1,
 DECODE(runstatus,9,'Success','UnSuccess') AS status ,
 p.*
 FROM psprcsrqst p
 WHERE prcsname LIKE '%'
 ORDER BY prcsname,
 prcsinstance desc, status DESC;

标签： plsqlpeoplesoft