How to determine if average is 0 vs null? - mongodb

Question

I've got a Mongo database where I run some aggregation queries. Here's the simplified query I want to run:
db.coll.aggregate([
{ $group: {
_id: 'fieldA',
fieldB: { $avg: '$fieldB' }
} },
])
It groups data by fieldA and calculates average for fieldB. Anyway, some rows in result set have 0 as value for fieldB. There can be 2 reasons for that:
Average value IS 0.
All documents in a group didn't have fieldB (or had null as a value); and Mongo behavior is to return 0 in that case.
Is it possible to determine which scenario took place for each row in resulting selection without issuing other query and without leaving aggregation pipeline?
UPDATE
I can't filter out non-null fields, as I'm doing aggregation for few fields, like that:
db.coll.aggregate([
{ $group: {
_id: 'fieldA',
fieldB: { $avg: '$fieldB' },
fieldC: { $avg: '$fieldC' }
} },
])
Some of the documents may have fieldB but not fieldC and vice versa.

Answer 1

You can filter the data by using $match before your $group operation.
db.coll.aggregate([
{ $match: { fieldB : {$ne : null }}}},
{ $group: {
_id: 'fieldA',
fieldB: { $avg: '$fieldB' }
} },
])
This way you will get only documents that have fieldB set.
UPDATE
You can't use the $avg that way but you can find out if all values are NULL using $min operator:
db.coll.aggregate([
{ $group: {
_id: 'fieldA',
fieldB: { $avg: '$fieldB' } ,
fieldBAllNullOrMin: { $min: '$fieldB' }
} },
])
The $min operator will return null if all values are null, otherwise it will return min. value (but only in 2.4+ versions of MongoDB).

Answer 2

You can use the $max (or $min) operator to determine whether all
instances of fieldB in a group are null or missing, as the $max (or
$min) operator return null in that case. Given this aggregation
pipeline:
c.aggregate([
{$group: {
_id: '$fieldA',
avg: {$avg: '$fieldB'},
max: {$max: '$fieldB'},
}}
])
with these documents:
c.insert({fieldA: 1, fieldB: 3})
c.insert({fieldA: 1, fieldB: -3})
the result is:
{"_id": 1, "avg": 0, "max": 3}
whereas with these documents:
c.insert({fieldA: 1})
c.insert({fieldA: 1})
the result is:
{"_id": 1, "avg": 0, "max": null}
The null value for the max field tells you that fieldB was null or
missing in all documents in the group.
Hope this helps,
Bruce