MongoDB nested grouping - mongodb

Question

I have the following MongoDB data model:
{
"_id" : ObjectId("53725814740fd6d2ee0ca2bb"),
"date" : "2014-01-01",
"establishmentId" : 1,
"products" : [
{
"productId" : 1,
"price" : 7.03,
"someOtherInfo" : 325,
"somethingElse" : 6878
},
{
"productId" : 2,
"price" : 4.6,
"someOtherInfo" : 243,
"somethingElse" : 1757
},
{
"productId" : 3,
"price" : 2.14,
"someOtherInfo" : 610,
"somethingElse" : 5435
},
{
"productId" : 4,
"price" : 1.45,
"someOtherInfo" : 627,
"somethingElse" : 5762
},
{
"productId" : 5,
"price" : 3.9,
"someOtherInfo" : 989,
"somethingElse" : 3752
}
}
What is the fastest way to get the average price across all establishments? Is there a better data model to achieve this?

Answer 1

An aggregation operation should handle this well. I'd suggest looking into the $unwind operation.
Something along these lines should work (just as an example):
db.collection.aggregate(
{$match: {<query parameters>}},
{$unwind: "$products"},
{
$group: {
_id: "<blank or field(s) to group by before averaging>",
$avg: "$price"
}
}
);
An aggregation built in this style should produce a JSON object that has the data you want.

Answer 2

Due to the gross syntax errors in anything else provided the more direct answer is:
db.collection.aggregate([
{ "$unwind": "$products" },
{ "$group": {
"_id": null,
"avgprice": { "$avg": "$products.price" }
}}
])
The usage of the aggregation framework here is to first $unwind the array, which is a way to "de-normalize" the content in the array into separate documents.
Then in the $group stage you pass in a value of null to the _id which means "group everything" and pass your $products.price ( note the dot notation ) in to the $avg operator to return the total average value across all of the sub-document entries in all of your documents in the collection.
See the full operator reference for more information.

Answer 3

The best solution I found was:
db.collection.aggregate([
{$match:{date:{$gte:"2014-01-01",$lte:"2014-01-31"},establishmentId:{$in:[1,2,3,4,5,6]}}
{ "$unwind": "$products" },
{ "$group": {
"_id": {date:"$date",product:"$products.productId"},
"avgprice": { "$avg": "$products.price" }
}}
])
And something I found out also was that it is much better to first use match and then unwind so there are fewer items to unwind. This results in a faster overall process.