MongoDB aggregation, Group by value interval, - mongodb

Question

MongoDB documents:
[{
_id: '123213',
elevation: 2300,
area: 25
},
{
_id: '343221',
elevation: 1600,
area: 35,
},
{
_id: '545322',
elevation: 500
area: 12,
},
{
_id: '234234',
elevation: null,
area: 5
}]
I want to group these on a given interval on elevation and summarize the area property.
Group 1: < 0
Group 2: 0 - 1500
Group 3: 1501 - 3000,
Group 4: > 3000
So the expected output would be:
[{
interval: '1501-3000',
count: 2,
summarizedArea: 60
},
{
interval: '0-1500',
count: 1,
summarizedArea: 12,
},
{
interval: 'N/A',
count: 1,
summarizedArea: 5
}]
If possible, I want to use the aggregation pipeline.
Maybe something with $range? Or a combination of $gte and $lte?

Answer 1

As Feliix suggested $bucket should do the job, but boundaries should be slightly different to play well with negative and N/A values:
db.collection.aggregate([
{
$bucket: {
groupBy: "$elevation",
boundaries: [ -Number.MAX_VALUE, 0, 1501, 3001, Number.POSITIVE_INFINITY ],
default: Number.NEGATIVE_INFINITY,
output: {
"count": { $sum: 1 },
"summarizedArea" : { $sum: "$area" }
}
}
}
])
The formatting stage below can be added to the pipeline to adjust shape of the response:
{ $group: {
_id: null,
documents: { $push: {
interval: { $let: {
vars: {
idx: { $switch: {
branches: [
{ case: { $eq: [ "$_id", -Number.MAX_VALUE ] }, then: 3 },
{ case: { $eq: [ "$_id", 0 ] }, then: 2 },
{ case: { $eq: [ "$_id", 1501 ] }, then: 1 },
{ case: { $eq: [ "$_id", 3001 ] }, then: 0 }
],
default: 4
} }
},
in: { $arrayElemAt: [ [ ">3000", "1501-3000", "0-1500", "<0", "N/A" ], "$$idx" ] }
} },
count: "$count",
summarizedArea: "$summarizedArea"
} }
} }
$group with _id: null $push es all groups into array of a single document.
$let maps $_id from previous stage to text labels of interval defined in the array [ ">3000", "1501-3000", "0-1500", "<0", "N/A" ]. For that it calculates idx index of the label using $switch.
It must be way simpler to implement the logic on application level unless you absolutely need to do it in the pipeline.

Answer 2

you can use $bucket introduced in MongoDB 3.4 to achive this:
db.collection.aggregate([
{
$bucket: {
groupBy: "$elevation",
boundaries: [
0,
1500,
3000,
5000
],
default: 10000,
output: {
"count": {
$sum: 1
},
"summarizedArea": {
$sum: "$area"
}
}
}
}
])
output:
[
{
"_id": 0,
"count": 1,
"summarizedArea": 12
},
{
"_id": 1500,
"count": 2,
"summarizedArea": 60
},
{
"_id": 10000,
"count": 1,
"summarizedArea": 5
}
]
you can try it here: mongoplayground.net/p/xFe7ZygMqaY