wrong result in MongoDB mapreduce function? - mongodb

Question

I have Collection "cars" from that want to get count of certified cars as trueCount and flaseCount where certified is boolean.
am issuing the following mapreduce query
map:-
function() { for (var idx = 0; idx < this.cars.length; idx++) {
var key = this.cars[idx].carName;
var value = {
count : 1,
certifiedCheck : this.cars[idx].certified
};
emit(key, value);
} }
reduce:-
function(key, values) {
certifiedCount = { trueCount: 0, falseCount: 0 };
values.forEach(function(value) {
if ( value.certifiedCheck )
certifiedCount.trueCount += value.count;
else
certifiedCount.falseCount += value.count;
});
return certifiedCount;
query:
{ "type": "cars" }
getting the following result :
{ "id" : "carName" , "value" : { "true" : 277.0 , "false" : NaN}};
even though I have 457 documents in the collection.
Please someone help me here to fix this issue.
Thanks in advance

Answer 1

You mixed up your map-reduce: to reduce to two keys "true" and "false" you need to emit these as keys. Then, the reducer will run per key.
As pseudo code:
map:
for each car
evaluate whether it should be true or false
key = (true/false)
emit(key, { count : 1 })
reduce:
(input is true/false as key, array of count-documents as value)
for each value-document
sum up the count
return key, sum
This should yields two documents with true / false as key and the respective sum as value.

Answer 2

You should consider using the aggregation framework for running the aggregation since it achieves the same result albeit faster than MapReduce as aggregation runs natively in the server (C++), MapReduce spawns separate javascript thread(s) to run JS code.
Thus said, if you run the following aggregation pipeline which uses the $cond operator to evaluate the counts based on the logic in the field expression, you will get a similar result:
Because you haven't showed your collection schema, I've assumed the following sample documents with a cars field as array having seen in your mapReduce you are doing a for loop on the cars property:
Populate test collection
db.collection.insert([
{ _id: 1, cars: [ { model: "A", certified: true }, { model: "B", certified: true } ] },
{ _id: 2, cars: [ { model: "A", certified: false }, { model: "B", certified: true } ] },
{ _id: 3, cars: [ { model: "A", certified: true }, { model: "B", certified: false } ] },
{ _id: 4, cars: [ { model: "A", certified: true }, { model: "B", certified: false } ] },
{ _id: 5, cars: [ { model: "A", certified: true }, { model: "B", certified: true } ] }
])
Run aggregation operation
db.collection.aggregate([
{ "$unwind": "$cars" },
{
"$group": {
"_id": "$cars.model",
"trueCount": {
"$sum": {
"$cond": [ "$cars.certified", 1, 0 ]
}
},
"falseCount": {
"$sum": {
"$cond": [ "$cars.certified", 0, 1 ]
}
}
}
}
])
Result:
/* 1 */
{
"_id" : "A",
"trueCount" : 4,
"falseCount" : 1
}
/* 2 */
{
"_id" : "B",
"trueCount" : 3,
"falseCount" : 2
}