在PostgreSQL中，如何计算其列值必须大于平均值的25％的列的实际平均值

Question

I've got 1 year data with million rows in a PostgreSQL 9.2 table, and I'm struggling to calculate one measure used for network statistics.

I have a table with columns date , devicename , traffic_rate

I need to find average traffic_rate excluding holidays . A day will be considered holiday if is traffic_rate is < 25% of average(traffic_rate) .Again I need to sum the traffic_rate by excluding holidays and find actual average . For eg:

date  devicename    traffic_rate
day1  gateway1        45.3
day2  gateway1        1
day3  gateway1        28.2
day4  gateway1        4.56

Average including holidays = (45.3+1+28.2+4.56)/4 = 19.76 Days with traffic_rate < 25 % of 19.76 will be treated as holiday --> Here day2,day4 is treated as holiday

Original traffic rate by excluding holidays = (45.3+28.2)/2 = 36.75

How can i implement this in postgresql query??

Any help will be highly appreciable

Answer 1

You could use a common table expression to select the global average first and then use this value to filter the table for the actual query:

WITH glob AS ( SELECT 0.25 * AVG( traffic_rate ) as reducedAverage FROM yourTab )
SELECT AVG( yourTab.traffic_rate )
FROM yourTab
INNER JOIN glob
  ON yourTab.traffic_rate >= glob.reducedAverage

Example Fiddle

Answer 2

Try this:

SELECT 
    AVG(traffic_rate) 
FROM traffic_tbl
WHERE traffic_rate > (SELECT 0.25 * AVG(traffic_rate) FROM traffic_tbl)

Answer 3

Here's another method:

SELECT
  AVG(traffic_rate)
FROM
(
  SELECT
    traffic_rate,
    AVG(traffic_rate) OVER () AS avg_traffic_rate
  FROM atable
) AS s
WHERE traffic_rate < 0.25 * avg_traffic_rate
;

This solution uses window aggregation ( AVG with an OVER clause). The s derived table contains individual traffic_rate values together with the average value calculated off the entire table. The main query is then uses the average value to filter out rows according to the requirement and calculates the final average based on the remaining values.