convert milliseconds to date in mongodb aggregation pipeline for group by?

I'm trying to get the logic behind converting the txnTime field to a date object because grouping by either a date field or a timestamp in milliseconds (like what you are presently doing) will yield the same result as they both are unique in their respective formats!

To change the txnTime field to a date object you should then include a $project pipeline before the $group pipeline stage with this expression

"txnTime": {
    "$add": [ new Date(0), "$txnTime" ]
}

so that you can do your $group operation on the converted/projected txnTime field:

var convertedTxnTime = { "$add": [new Date(0), "$txnTime"] };

/*
  If using MongoDB 4.0 and newer, use $toDate 

  var convertedTxnTime = { "$toDate": "$txnTime" };

  or $convert

  var convertedTxnTime = { "$convert": { "input": "$txnTime", "to": "date" } };

*/

db.campaign_wallet.aggregate([
    { "$match": { 
        "campaignId" : 1 , 
        "txnTime" : { 
            "$gte" : 1429554600000 , 
            "$lte" : 1430159400000
        }
    } },
    { "$group" : { 
        "_id" : {
            "txnTime": convertedTxnTime,
            "msisdn" : "$msisdn"
        }, 
        "msisdnCount" : { "$sum" : 1}
    } }
]);

Output: (based on the sample documents from this question)

/* 0 */
{
    "result" : [ 
        {
            "_id" : {
                "txnTime" : ISODate("2015-04-25T18:30:00.000Z"),
                "msisdn" : "91808770101"
            },
            "msisdnCount" : 1
        }, 
        {
            "_id" : {
                "txnTime" : ISODate("2015-04-27T05:11:54.796Z"),
                "msisdn" : "9180877010"
            },
            "msisdnCount" : 1
        }, 
        {
            "_id" : {
                "txnTime" : ISODate("2015-04-25T18:30:01.111Z"),
                "msisdn" : "91808070101"
            },
            "msisdnCount" : 1
        }, 
        {
            "_id" : {
                "txnTime" : ISODate("2015-04-25T18:30:00.000Z"),
                "msisdn" : "91808070101"
            },
            "msisdnCount" : 2
        }, 
        {
            "_id" : {
                "txnTime" : ISODate("2015-04-27T05:11:54.796Z"),
                "msisdn" : "9189877000"
            },
            "msisdnCount" : 1
        }, 
        {
            "_id" : {
                "txnTime" : ISODate("2015-04-27T05:11:54.796Z"),
                "msisdn" : "9189877667"
            },
            "msisdnCount" : 1
        }
    ],
    "ok" : 1
}

-- UPDATE --

To group the documents by date with the format YYYY-MM-DD, use the Date Aggregation Operators

Example:

var convertedTxnTime = { "$add": [new Date(0), "$txnTime"] };

/*
  If using MongoDB 4.0 and newer, use $toDate 

  var convertedTxnTime = { "$toDate": "$txnTime" };

  or $convert

  var convertedTxnTime = { "$convert": { "input": "$txnTime", "to": "date" } };

*/

db.campaign_wallet.aggregate([
    { "$match": { 
        "campaignId" : 1 , 
        "txnTime" : { 
            "$gte" : 1429554600000 , 
            "$lte" : 1430159400000
        }
    } },
    { "$group" : { 
        "_id" : {
            "txnTime_year" : { "$year": convertedTxnTime },
            "txnTime_month" : { "$month": convertedTxnTime },
            "txnTime_day" : { "$dayOfMonth": convertedTxnTime },
            "msisdn": "$msisdn"
        }, 
        "msisdnCount" : { "$sum" : 1}
    } }
]);

Output:

/* 0 */
{
    "result" : [ 
        {
            "_id" : {
                "txnTime_year" : 2015,
                "txnTime_month" : 4,
                "txnTime_day" : 25,
                "msisdn" : "91808770101"
            },
            "msisdnCount" : 1
        }, 
        {
            "_id" : {
                "txnTime_year" : 2015,
                "txnTime_month" : 4,
                "txnTime_day" : 25,
                "msisdn" : "91808070101"
            },
            "msisdnCount" : 3
        }, 
        {
            "_id" : {
                "txnTime_year" : 2015,
                "txnTime_month" : 4,
                "txnTime_day" : 27,
                "msisdn" : "9180877010"
            },
            "msisdnCount" : 1
        }, 
        {
            "_id" : {
                "txnTime_year" : 2015,
                "txnTime_month" : 4,
                "txnTime_day" : 27,
                "msisdn" : "9189877000"
            },
            "msisdnCount" : 1
        }, 
        {
            "_id" : {
                "txnTime_year" : 2015,
                "txnTime_month" : 4,
                "txnTime_day" : 27,
                "msisdn" : "9189877667"
            },
            "msisdnCount" : 1
        }
    ],
    "ok" : 1
}

With mongodb 4.0 you can try $toDate aggregation to convert milliseconds to date format

db.collection.aggregate([
  { "$match": { 
    "campaignId" : 1 , 
    "txnTime" : { 
      "$gte" : 1429554600000 , 
      "$lte" : 1430159400000
    }
  }},
  { "$project": {
    "toDate": {
      "$toDate": "$txnTime"
    }
  }}
])

You can try it here