In the elastic documentation it says I can perform multi-term aggregation if I use script (reference). It is not clear to me how this is done. Basically what I am after is count(*) ... group by logsource,pid. Without a script, it seems I can only do one group by.
Can someone show me an example?
Using script can be costly, but to answer your question,
POST /_search
{
"size": 0,
"aggs": {
"test": {
"terms": {
"script": "doc['logsource'].value+\":\"+doc['pid'].value",
"size": 0
}
}
}
}
Will do!
I think by using sub aggregations I can get the intended result, take for example:
{
"query" : {
"match": {
"message": "error"
}
},
"aggs": {
"g_logsource": {
"terms": {
"field": "logsource"
},
"aggs": {
"g_pid": {
"terms": {
"field": "pid"
},
"aggs" : {
"ts" : {
"date_histogram" : {
"field" : "#timestamp",
"interval" : "1h"
}
}
}
}
}
}
}
}
Returns:
"aggregations": {
"g_logsource": {
"doc_count_error_upper_bound": 0,
"buckets": [
{
"key": "nyhq",
"doc_count": 2129,
"g_pid": {
"doc_count_error_upper_bound": 5,
"buckets": [
{
"key": "5641",
"doc_count": 9,
"ts": {
"buckets": [
{
"key_as_string": "2014-12-07T04:00:00.000Z",
"key": 1417924800000,
"doc_count": 2
},
{
"key_as_string": "2014-12-07T08:00:00.000Z",
"key": 1417939200000,
"doc_count": 4
},
{
"key_as_string": "2014-12-07T18:00:00.000Z",
"key": 1417975200000,
"doc_count": 1
},
{
"key_as_string": "2014-12-07T20:00:00.000Z",
"key": 1417982400000,
"doc_count": 2
}
]
}
},
{
"key": "14839",
"doc_count": 3,
"ts": {
"buckets": [
{
"key_as_string": "2014-12-07T09:00:00.000Z",
"key": 1417942800000,
"doc_count": 1
},
{
"key_as_string": "2014-12-07T20:00:00.000Z",
"key": 1417982400000,
"doc_count": 2
}
]
}
}
In my code, I can then combine groups to be {logsource: nyhq, pid: 5641} as the identifer for each time series. I think this is the same as GROUP BY in SQL. Would appreciate any comments confirming this.
Related
I would like to find the difference between the min date of a series of buckets and the the date of that bucket. For instance I have an elastic aggregation similar to below
"id": {
"terms": {
"field": "data.id"
},
"aggs": {
"min_date": {
"min": {
"field": "data.dateSold"
}
},
"date": {
"date_histogram": {
"field": "data.dateSold",
"calendar_interval": "year"
},
"aggs": {
"sales": {
"sum": {
"field": "data.sales"
}
}
}
}
}
}
}
this produces a result similar too
{
"uwi": {
"doc_count_error_upper_bound": 0,
"sum_other_doc_count": 8559203,
"buckets": [
{
"key": "Tshirts",
"doc_count": 1826,
"date_histogram#date": {
"buckets": [
{
"key_as_string": "2021-01-01T00:00:00.000Z",
"key": 1609459200000,
"doc_count": 364,
"sum#sales": {
"value": 31438.67796
}
},
{
"key_as_string": "2022-01-01T00:00:00.000Z",
"key": 1640995200000,
"doc_count": 365,
"sum#sales": {
"value": 16095.7913
}
}
]
},
"min#min_date": {
"value": 1609459200000,
"value_as_string": "2021-01-01T00:00:00.000Z"
}
...
...
I would like to add an extra value per bucket that is the difference between the date (key) and the min date e.g. I get a result similar to below with an extra 'difference_with_min_date' value per bucket which is the diff between the 'min#min_date' agg and that buckets 'key'
{
"uwi": {
"doc_count_error_upper_bound": 0,
"sum_other_doc_count": 8559203,
"buckets": [
{
"key": "Tshirts",
"doc_count": 1826,
"date_histogram#date": {
"buckets": [
{
"key_as_string": "2021-01-01T00:00:00.000Z",
"key": 1609459200000,
"doc_count": 364,
sum#sales": {
"value": 31438.67796
},
"difference_with_min_date": {
"value": 0
}
},
{
"key_as_string": "2022-01-01T00:00:00.000Z",
"key": 1640995200000,
"doc_count": 365,
"sum#sales": {
"value": 16095.7913
},
"difference_with_min_date": {
"value": 31536000000
}
}
]
},
"min#min_date": {
"value": 1609459200000,
"value_as_string": "2021-01-01T00:00:00.000Z"
}
...
...
Any ideas would be helpful, I have tried to do this with a script with little success as you need to supply the bucket_script path within the 'sales' aggs (ie as a sibling) to do it per bucket value but then you cant reference the uncle min aggregation.
Thanks
I made query result that aggregate some data, and its aggregation key is number. I tried to sort result of aggregation by key. elasticsearch treated key as string.
Since the number of current result bucket is pretty large, it's unable to modify on client side. Any idea of this?
Here is my query.
"aggregations" : {
"startcount" : {
"terms" : {
"script" : "round(doc['startat'].value/1000)",
"size" : 1000,
"order" : { "_term" : "asc" }
}
}
}
and current result bucket.
"buckets": [
{
"key": "0",
"doc_count": 68
},
{
"key": "1",
"doc_count": 21
},
{
"key": "10",
"doc_count": 6
},
{
"key": "11",
"doc_count": 16
},
It's my expect result.
"buckets": [
{
"key": "0",
"doc_count": 68
},
{
"key": "1",
"doc_count": 21
},
{
"key": "2", // not '10'
"doc_count": 6
},
{
"key": "3", // not '11'
"doc_count": 16
},
Using the value_script approach should fix the alphabetical sort issue:
Example:
{
"size": 0,
"aggregations": {
"startcount": {
"terms": {
"field": "startat",
"script": "round(_value/1000)",
"size": 1000,
"order": {
"_term": "asc"
}
}
}
}
}
This is a multiple group by scenario where data are being sorted by the key descending order.
{
"size": 0,
"aggs": {
"categories": {
"filter": {
"exists": {
"field": "organization_industries"
}
},
"aggs": {
"names": {
"terms": {
"field": "organization_revenue_in_thousands_int.keyword",
"size": 200,
"order": {
"_key": "desc"
}
},
"aggs": {
"industry_stats": {
"terms": {
"field": "organization_industries.keyword"
}
}
}
}
}
}
}
}
Output
"aggregations": {
"categories": {
"doc_count": 195161605,
"names": {
"doc_count_error_upper_bound": 0,
"sum_other_doc_count": 19226983,
"buckets": [
{
"key": "99900",
"doc_count": 1742,
"industry_stats": {
"doc_count_error_upper_bound": 0,
"sum_other_doc_count": 0,
"buckets": [
{
"key": "internet",
"doc_count": 1605
},
{
"key": "investment management",
"doc_count": 81
},
{
"key": "biotechnology",
"doc_count": 54
},
{
"key": "computer & network security",
"doc_count": 2
}
]
}
},
{
"key": "998000",
"doc_count": 71,
"industry_stats": {
"doc_count_error_upper_bound": 0,
"sum_other_doc_count": 0,
"buckets": [
{
"key": "finance",
"doc_count": 48
},
{
"key": "information technology & services",
"doc_count": 23
}
]
}
}
}
]
}
}
enter code here
I have a json data in the below format
{
"ID": { "Color": "Black", "Product": "Car" },
"ID": { "Color": "Black", "Product": "Car" },
"ID": { "Color": "Black", "Product": "Van" },
"ID": { "Color": "Black", "Product": "Van" },
"ID": { "Color": "Ash", "Product": "Bike" }
}
I want to calculate the count of car and the corresponding color. I am using elasticsearch facet to do this.
My query
$http.post('http://localhost:9200/product/productinfoinfo/_search?size=5', { "aggregations": { "ProductInfo": { "terms": { "field": "product" } } }, "facets": { "ProductColor": { "terms": { "field": "Color", "size": 10 } } } })
I am getting the output like below
"facets": { "ProductColor": { "_type": "terms", "missing": 0, "total": 7115, "other": 1448, "terms": [ { "term": "Black", "count": 4 }, { "term": "Ash","count":1} },
"aggregations": { "ProductInfo": { "doc_count_error_upper_bound": 94, "sum_other_doc_count": 11414, "buckets": [ { "key": "Car", "doc_count": 2 }, { "key": "Van", "doc_count": 2 }, { "key": "Bike", "doc_count": 1 } ] } } }
What I actually want is,
[ { "key": "Car", "doc_count": 2, "Color":"Black", "count":2 }, { "key": "Van", "doc_count": 2,"Color":"Black", "count":2 }, { "key": "Bike", "doc_count": 1,"Color":"Ash", "count":1 } ]
I would like to groupby the result . Is it possible to do it in elasticsearch query.
Thanks in advance
This is because you're using both aggregations and facets, which, if they are similar, are not meant to be used together.
Facets are deprecated and will be soon removed from ElasticSearch.
Aggregations are the way to go to make "group by"-like queries.
You just have to nest another terms aggregation in the first one, like this :
{
"aggs": {
"By_type": {
"terms": {
"field": "Product"
},
"aggs": {
"By_color": {
"terms": {
"field": "Color"
}
}
}
}
}
}
And the result will be close to what you want :
"aggregations": {
"By_type": {
"doc_count_error_upper_bound": 0,
"sum_other_doc_count": 0,
"buckets": [
{
"key": "bike",
"doc_count": 2,
"By_color": {
"doc_count_error_upper_bound": 0,
"sum_other_doc_count": 0,
"buckets": [
{
"key": "ash",
"doc_count": 1
},
{
"key": "black",
"doc_count": 1
}
]
}
},
{
"key": "car",
"doc_count": 2,
"By_color": {
"doc_count_error_upper_bound": 0,
"sum_other_doc_count": 0,
"buckets": [
{
"key": "black",
"doc_count": 2
}
]
}
},
{
"key": "van",
"doc_count": 1,
"By_color": {
"doc_count_error_upper_bound": 0,
"sum_other_doc_count": 0,
"buckets": [
{
"key": "black",
"doc_count": 1
}
]
}
}
]
}
}
here is the data in my elasticsearch server:
{"system": "aaa"},
{"system": "bbb"},
{"system": null}
I want to get the statistics for system. then I did the query:
{
"aggs" : {
"myAggrs" : {
"terms" : { "field" : "system" }
}
}
it gives me the result:
{
"key": "aaa",
"doc_count": 1
},
{
"key": "bbb",
"doc_count": 1
}
but the "key" : null is not included in the result, how can I get it?
here is my expect result:
{
"key": "aaa",
"doc_count": 1
},
{
"key": "bbb",
"doc_count": 1
},
{
"key": null,
"doc_count": 1
}
I don't think you can do this with terms. Try with another aggregation:
{
"aggs": {
"myAggrs": {
"terms": {
"field": "system"
}
},
"missing_system": {
"missing": {
"field": "system"
}
}
}
}
And the result will be:
"aggregations": {
"myAggrs": {
"doc_count_error_upper_bound": 0,
"sum_other_doc_count": 0,
"buckets": [
{
"key": "aaa",
"doc_count": 1
},
{
"key": "bbb",
"doc_count": 1
}
]
},
"missing_system": {
"doc_count": 1
}
}
I have the following aggregation for Categories
{
"aggs": {
"category": {
"terms": { "field": "category.name" }
}
}
}
// results
"category": {
"buckets": [
{
"key": "computer & office",
"doc_count": 365
},
{
"key": "home & garden",
"doc_count": 171
},
{
"key": "consumer electronics",
"doc_count": 49
},
]
}
How can I pass additional field, like category.id to the category buckets, so I could query by category.id if the certain aggregation is clicked by a user. I'm not really clear how to query aggregations, if there's any direct way or you have to make a new query and pass bucket key to query filters.
Use a sub-aggregation on the category.id, you will do a bit more work when looking at the results, but I think it's better than changing the mapping:
{
"aggs": {
"name": {
"terms": {
"field": "name"
},
"aggs": {
"id": {
"terms": {
"field": "id"
}
}
}
}
}
}
And the results will look like the following:
"aggregations": {
"name": {
"doc_count_error_upper_bound": 0,
"sum_other_doc_count": 0,
"buckets": [
{
"key": "consumer electronics",
"doc_count": 2,
"id": {
"doc_count_error_upper_bound": 0,
"sum_other_doc_count": 0,
"buckets": [
{
"key": 2,
"doc_count": 2
}
]
}
},
{
"key": "computer & office",
"doc_count": 1,
"id": {
"doc_count_error_upper_bound": 0,
"sum_other_doc_count": 0,
"buckets": [
{
"key": 5,
"doc_count": 1
}
]
}
},
{
"key": "home & garden",
"doc_count": 1,
"id": {
"doc_count_error_upper_bound": 0,
"sum_other_doc_count": 0,
"buckets": [
{
"key": 1,
"doc_count": 1
}
]
}
},
{
"key": "whatever",
"doc_count": 1,
"id": {
"doc_count_error_upper_bound": 0,
"sum_other_doc_count": 0,
"buckets": [
{
"key": 3,
"doc_count": 1
}
]
}
}
]
}
}
You will still have the category name, but now you, also, have the id from the second aggregation as a sub-bucket in the root bucket:
"key": "consumer electronics",
...
"id": {
...
"buckets": [
{
"key": 2,
"doc_count": 2
You could add a sub aggregation:
{
"aggs": {
"category": {
"terms": {
field": "category.name",
"aggs": {
"id": {
"terms": { "field": "category.id" }
}
}
}
}
}
}
This way each category.name bucket will contain a single bucket containing the id for that category.