Hi,
I'm trying to calculate a serial difference between 2 values in different documents that are related by a project name and a "voie" fields (first part of the query).
Here is where i am :
{
"query": {
"bool": {
"must": [
{ "match": { "nomProjet": "myProject2"} },
{ "match": { "voie": "3"} }
]
}
},
"aggs": {
"temp": {
"histogram": {
"field": "timePaquet",
"interval": "0.1"
},
"aggs": {
"monthly": {
"sum": {
"field": "mean"
}
}
}
},
"sum_bucket_demo": {
"serial_diff": {
"buckets_path": "temp>monthly",
"lag": 1
}
}
}
}
As you can see, I don't have a timestamp field but I do have a ascendant numeric value (timePaquet). that's why I use a histogram instead of a date_histogram.
After a lot of comparisons with the documentation for serial_diff in elastic, I can't find my mistake.
The current error reason is : "Invalid pipeline aggregation named [sum_bucket_demo] of type [serial_diff]. Only sibling pipeline aggregations are allowed at the top level"
note : it's actually working with the sum_bucket aggregation.
Any advice would be much appreciated