Elasticsearch的metrics(度量)包含count、sum、avg、max、min、percentiles(百分位数)、Unique count(基数 || 去重计数)、Median(中位数)、扩展度量(含方差、平方和、标准差、标准差界限)、Percentile ranks(百分位等级)
1.count(数量):
SELECT count(log_date.d) AS Count FROM INDEX-2017-12
2.sum(和):
SELECT sum(log_date.d) AS SUM FROM INDEX-2017-12
3.avg(平均数):
SELECT avg(log_date.d) AS AVG FROM INDEX-2017-12
4.max(最大值):
SELECT max(log_date.d) AS MAX FROM INDEX-2017-12
5.min(最小值):
SELECT min(log_date.d) AS MIN FROM INDEX-2017-12
6.percentiles(百分位数):
SELECT percentiles(log_date.d,1.0,15.0,31.0) AS Percentiles FROM INDEX-2017-12
7.Unique count(基数 || 去重计数,就是SQL中的distinct):
SELECT count(distinct(log_date.d)) AS UniqueCount FROM INDEX-2017-12
8.Median(中位数):
中位数没找到单独的获取方法,不过在Kibana中看到获取中位数时请求中的参数,其实就是获取的某个字段50的百分位数,所以可能有:中位数=50的百分位数
SELECT percentiles(log_date.d,50.0) AS percentiles FROM INDEX-2017-12
9.方差、平方和、标准差、标准差界限:
这几个度量没有单独方法去获取,都是用EXTENDED_STATS一个请求全部获取下来,然后从中取自己需要的结果
SELECT EXTENDED_STATS(log_date.d) AS EXTENDED_STATS FROM INDEX-2017-12
EXTENDED_STATS查询结果包含:方差、平方和、标准差、标准差界限以及最大值、平均数等基础度量,具体如下:
"aggregations": { "1": { "count": 15304326, "min": 1, "max": 31, "avg": 15.068216202399244, "sum": 230608893, "sum_of_squares": 4588588661, "variance": 72.7718426201877, "std_deviation": 8.530641395591992, "std_deviation_bounds": { "upper": 32.129498993583226, "lower": -1.9930665887847407 } } }
10.Percentile ranks(百分位等级)
暂时没找到求百分位等级的SQL语句,只能用原生ES查询语句获取了;
ES原生查询语句如下:
{ "size": 0, ...... "aggs": { "1": { "percentile_ranks": { "field": "log_date.d", "values": [ 6, 15, 31 ], "keyed": false } } } }
附elasticsearch-sql的GitHub地址:https://github.com/NLPchina/elasticsearch-sql
Elasticsearch官方文档(中文版)地址:https://www.elastic.co/guide/cn/elasticsearch/guide/cn/aggregations.html