Averaging across list indicies during aggregation pipeline? - mongodb

Question

I currently have a MongoDB aggregation pipeline that ends with the following type of synthetic documents
[
{
'_id': '2019-09-10',
'grouped_foos':
[
{... 'foo': [1, 78, 100]},
{... 'foo': [8, 66, 98]},
{... 'foo': [99, 5, 33]},
{... 'foo': [120, 32, 2]}
]
},
{
'_id': '2019-09-09',
'grouped_foos':
[
{... 'foo': [10, 27]},
{... 'foo': [19, 66]}
]
},
{
'_id': '2019-09-08',
'grouped_foos':
[
{... 'foo': [1]}
]
}
]
I would like to continue this pipeline and average the indices of the foo lists together to form documents that look like
[
{
'_id': '2019-09-10',
'avg_foo': [57, 45.25, 58.25]
},
{
'_id': '2019-09-09',
'avg_foo': [14.5, 46.5]
},
{
'_id': '2019-09-08',
'avg_foo': [1]
}
]
Is this type of averaging possible during aggregation? Do I potentially need to $unwind the lists with indexing and assign new _id for uniqueness to make documents that look like
[
{
'_id': UUID,
'old_id': '2019-09-10',
'foo': 1,
'index': 0
},
{
'_id': UUID,
'old_id': '2019-09-10',
'foo': 78,
'index': 1
},
........
]

Answer 1

Basically you can try with $unwind but easier and faster approach would be to use $reduce to $map and $sum all the rows from grouped_foos. Then you'll be able to run another $map and use $divide to get the average.
db.collection.aggregate([
{
$project: {
size: { $size: "$grouped_foos" },
foo_sum: {
$reduce: {
input: "$grouped_foos",
initialValue: [],
in: {
$map: {
input: { $range: [ 0, { $size: "$$this.foo" }, 1 ] },
as: "index",
in: {
$add: [
{ $arrayElemAt: [ "$$this.foo", "$$index" ] },
{ $ifNull: [ { $arrayElemAt: [ "$$value", "$$index" ] }, 0 ] }
]
}
}
}
}
}
}
},
{
$project: {
_id: 1,
avg_foo: {
$map: {
input: "$foo_sum",
in: {
$divide: [ "$$this", "$size" ]
}
}
}
}
}
])
Mongo Playground