How to arrive at the difference in values for minimum and maximum dates for each year in mongodb - mongodb-query

Question

How to arrive at the difference in values of a specific column for the records with minimum and maximum dates for each year in a collection using the aggregate pipeline or map reduce in mongodb?
I have the following collection:
/* 1 */
{
"_id" : 1,
"item" : "abc",
"price" : 10,
"quantity" : 2,
"date" : ISODate("2014-01-01T08:00:00.000Z")
},
/* 2 */
{
"_id" : 2,
"item" : "jkl",
"price" : 20,
"quantity" : 1,
"date" : ISODate("2014-02-03T09:00:00.000Z")
},
/* 3 */
{
"_id" : 3,
"item" : "xyz",
"price" : 5,
"quantity" : 5,
"date" : ISODate("2014-02-03T09:05:00.000Z")
},
/* 4 */
{
"_id" : 4,
"item" : "abc",
"price" : 10,
"quantity" : 10,
"date" : ISODate("2014-02-15T08:00:00.000Z")
},
/* 5 */
{
"_id" : 5,
"item" : "xyz",
"price" : 5,
"quantity" : 10,
"date" : ISODate("2014-02-15T09:05:00.000Z")
},
/* 6 */
{
"_id" : 6,
"item" : "abc",
"price" : 10,
"quantity" : 2,
"date" : ISODate("2013-01-01T08:00:00.000Z")
},
/* 7 */
{
"_id" : 7,
"item" : "jkl",
"price" : 20,
"quantity" : 1,
"date" : ISODate("2013-02-03T09:00:00.000Z")
},
/* 8 */
{
"_id" : 8,
"item" : "xyz",
"price" : 5,
"quantity" : 5,
"date" : ISODate("2013-02-03T09:05:00.000Z")
},
/* 9 */
{
"_id" : 9,
"item" : "abc",
"price" : 10,
"quantity" : 10,
"date" : ISODate("2013-02-15T08:00:00.000Z")
},
/* 10 */
{
"_id" : 10,
"item" : "xyz",
"price" : 5,
"quantity" : 10,
"date" : ISODate("2013-02-15T09:05:00.000Z")
},
/* 11 */
{
"_id" : 11,
"item" : "abc",
"price" : 10,
"quantity" : 2,
"date" : ISODate("2012-01-01T08:00:00.000Z")
},
/* 12 */
{
"_id" : 12,
"item" : "jkl",
"price" : 20,
"quantity" : 1,
"date" : ISODate("2012-02-03T09:00:00.000Z")
},
/* 13 */
{
"_id" : 13,
"item" : "xyz",
"price" : 5,
"quantity" : 5,
"date" : ISODate("2012-02-03T09:05:00.000Z")
},
/* 14 */
{
"_id" : 14,
"item" : "abc",
"price" : 10,
"quantity" : 10,
"date" : ISODate("2012-02-15T08:00:00.000Z")
},
/* 15 */
{
"_id" : 15,
"item" : "xyz",
"price" : 5,
"quantity" : 10,
"date" : ISODate("2012-02-15T09:05:00.000Z")
},
I would like the result to take the following form:
{
{"year": 2014}, {"minDtQuantity": 2}, {"maxDtQuantity": 10}, {"quantityDiff": 8},
{"year": 2013}, {"minDtQuantity": 2}, {"maxDtQuantity": 10}, {"quantityDiff": 8},
{"year": 2012}, {"minDtQuantity": 2}, {"maxDtQuantity": 10}, {"quantityDiff": 8},
}
For each year, we need to find the minimum and maximum dates and group them by year and then find the "quantity" values on those dates and then find the difference between the quantities for the min and max dates for each year.
Is that even possible with aggregate pipelines or map-reduce in mongodb?

Answer 1

This can be done using aggregation pipelines by sorting by date, then pushing the quantities into arrays when grouping by the year (Use a the $year operator to extract the year from the date object). The quantities for the minimum and maximum dates in that year are then the first and last values in the array respectively. These can be taken out of the array using $arrayElemAt.
db.collection.aggregate(
[
{
$sort: {
"date": 1
}
},
{
$group: {
"_id": { "$year": "$date" },
"quantityArray": { "$push": "$quantity" },
}
},
{
$project: {
"_id": 0,
"year": "$_id",
"minDtQuantity": { "$arrayElemAt": [ "$quantityArray", 0 ] },
"maxDtQuantity": { "$arrayElemAt": [ { "$reverseArray": "$quantityArray" }, 0 ] },
"quantityDiff": { "$subtract": [
{ "$arrayElemAt": [ { "$reverseArray": "$quantityArray" }, 0 ] },
{ "$arrayElemAt": [ "$quantityArray", 0 ] },
] }
}
},
]
);
This aggregation returns these results on your data:
{
"year" : NumberInt(2014),
"minDtQuantity" : NumberInt(2),
"maxDtQuantity" : NumberInt(10),
"quantityDiff" : NumberInt(-8)
},
{
"year" : NumberInt(2013),
"minDtQuantity" : NumberInt(2),
"maxDtQuantity" : NumberInt(10),
"quantityDiff" : NumberInt(-8)
},
{
"year" : NumberInt(2012),
"minDtQuantity" : NumberInt(2),
"maxDtQuantity" : NumberInt(10),
"quantityDiff" : NumberInt(-8)
}
This is not quite the format you specified. I am not exactly sure what you required, did you need the results returned in one document?