ElasticSearch(七):聚合分析简介
ElasticSearch(七):聚合分析简介
## 什么是聚合(Aggregation) * ElasticSearch除了搜索外,还提供了针对ElasticSearch数据进行统计分析的功能 * 通过聚合,会得到一个数据的概览,这样就可以分析和总结全套的数据,而不是仅仅能寻找单个文档 * 易用性,只需要一条语句,就可以从ElasticSearch中得到分析结果,从而避免在客户端实现分析逻辑
## 聚合的分类 * Bucket Aggregation:一些列满足特定条件的文档集合,其相当于SQL中的Group By * Metric Aggregation:一些数学运算,可以对文档字段进行统计分析 - 其相当于SQL中对于字段进行运算的函数,如Sum、Count等。它除了可以在字段上进行计算,还可以在脚本产生的结果上进行计算 - 大多数Metric是数学计算,输出一个值,如:min/max/sum/avg/cardinality - 部分支持输出多个值,如:stats/percentiles/percentile_ranks * Pipeline Aggregation:对其他的聚合结果进行二次聚合 * Matrix Aggregation:支持对多个字段的操作并提供一个结果矩阵
## 一些聚合分析例子
#按照目的地进行分桶统计
GET kibana_sample_data_flights/_search
{
"size": 0,
"aggs":{
"flight_dest":{
"terms":{
"field":"DestCountry"
}
}
}
}
#查看航班目的地的统计信息,增加平均,最高最低价格
GET kibana_sample_data_flights/_search
{
"size": 0,
"aggs":{
"flight_dest":{
"terms":{
"field":"DestCountry"
},
"aggs":{
"avg_price":{
"avg":{
"field":"AvgTicketPrice"
}
},
"max_price":{
"max":{
"field":"AvgTicketPrice"
}
},
"min_price":{
"min":{
"field":"AvgTicketPrice"
}
}
}
}
}
}
#价格统计信息+天气信息
GET kibana_sample_data_flights/_search
{
"size": 0,
"aggs":{
"flight_dest":{
"terms":{
"field":"DestCountry"
},
"aggs":{
"stats_price":{
"stats":{
"field":"AvgTicketPrice"
}
},
"wather":{
"terms": {
"field": "DestWeather",
"size": 5
}
}
}
}
}
}
public boolean getJuelExpression(JSONObject condition, JSONObject formData) {
StringBuilder juelExpression = new StringBuilder();
ExpressionFactory factory = new ExpressionFactoryImpl();
SimpleContext context = new SimpleContext();
String logic = condition.getString("logic");
JSONArray groups = condition.getJSONArray("groups");
int groupSize = groups.size();
for (int i = 0; i < groupSize; i++) {
JSONObject group = groups.getJSONObject(i);
if (i == 0) {
juelExpression.append("(");
} else {
juelExpression.append(logic + "(");
}
String groupLogic = group.getString("logic");
JSONArray expressions = group.getJSONArray("expressions");
int expressionSize = expressions.size();
for (int j = 0; j < expressionSize; j++) {
JSONObject expression = expressions.getJSONObject(j);
String field = expression.getString("field");
String compare = expression.getString("compare");
String type = expression.getString("type");
if ("String".equals(type)) {
String valueStr = expression.getString("value");
//valueStrKey防止变量名重复
String valueStrKey = field + "ValueStr" + i + j;
if (j == 0) {
if (compare.equals("==")) {
juelExpression.append(field + ".equals(" + valueStrKey + ")");
} else if (compare.equals("!=")) {
juelExpression.append("!" + field + ".equals(" + valueStrKey + ")");
}
} else {
if (compare.equals("==")) {
juelExpression.append(groupLogic + field + ".equals(" + valueStrKey + ")");
} else if (compare.equals("!=")) {
juelExpression.append(groupLogic + "!" + field + ".equals(" + valueStrKey + ")");
}
}
context.setVariable(field, factory.createValueExpression(formData.getString(field), String.class));
context.setVariable(valueStrKey, factory.createValueExpression(valueStr, String.class));
logger.info("field:{},fieldValue:{},valueStrKey:{},valueStr:{}", field, formData.getString(field), valueStrKey, valueStr);
} else if ("Number".equals(type)) {
long valueNum = expression.getLongValue("value");
//valueNumKey防止变量名重复
String valueNumKey = field + "ValueStr" + i + j;
if (j == 0) {
juelExpression.append(field + compare + valueNumKey);
} else {
juelExpression.append(groupLogic + field + compare + valueNumKey);
}
context.setVariable(field, factory.createValueExpression(formData.getLongValue(field), long.class));
context.setVariable(valueNumKey, factory.createValueExpression(valueNum, long.class));
logger.info("field:{},fieldValue:{},valueNumKey:{},valueNum:{}", field, formData.getLongValue(field), valueNumKey, valueNum);
} else {
//级联类型type为:Cascade
JSONArray valueArr = expression.getJSONArray("value");
//valueArrKey防止变量名重复
String valueArrKey = field + "ValueArr" + i + j;
if (j == 0) {
if (compare.equals(".contains")) {
juelExpression.append(field + ".containsAll(" + valueArrKey + ")");
} else if (compare.equals("!.contains")) {
juelExpression.append("!" + field + ".containsAll(" + valueArrKey + ")");
} else if (compare.equals("==")) {
juelExpression.append(field + ".equals(" + valueArrKey + ")");
} else if (compare.equals("!=")) {
juelExpression.append("!" + field + ".equals(" + valueArrKey + ")");
}
} else {
if (compare.equals(".contains")) {
juelExpression.append(groupLogic + field + ".containsAll(" + valueArrKey + ")");
} else if (compare.equals("!.contains")) {
juelExpression.append(groupLogic + "!" + field + ".containsAll(" + valueArrKey + ")");
} else if (compare.equals("==")) {
juelExpression.append(groupLogic + field + ".equals(" + valueArrKey + ")");
} else if (compare.equals("!=")) {
juelExpression.append(groupLogic + "!" + field + ".equals(" + valueArrKey + ")");
}
}
context.setVariable(field, factory.createValueExpression(formData.getJSONArray(field), JSONArray.class));
context.setVariable(valueArrKey, factory.createValueExpression(valueArr, JSONArray.class));
logger.info("field:{},fieldValue:{},valueArrKey:{},valueArr:{}", field, formData.getJSONArray(field), valueArrKey, valueArr);
}
}
juelExpression.append(")");
}
String lastJuel = "${" + juelExpression.toString() + "}";
logger.info("lastJuel:{} ", lastJuel);
ValueExpression e = factory.createValueExpression(context, lastJuel, boolean.class);
return (boolean) e.getValue(context);
}
作者:牧汜
出处:http://www.cnblogs.com/czbxdd/
本文版权归作者和博客园共有,欢迎转载,但未经作者同意必须保留此段声明,且在文章页面明显位置给出原文连接,否则保留追究法律责任的权利。