MongoDB: Calculate dwell time between every status value change

Question

I want to find out the dwell time between every presenceStatus change.

Example collection -

   /* 1 */

{
    \"_id\" : ObjectId(\"5e4889a7c7959f6a         


        

Answer 1

@Valijon, @Plancke Thanks all who helped. we chose to take a different approach and turned our a simple for loop was able to do the job for now. Thanks again. Here is final solution if any one was interested:

let prevSensingResults = {}; 

db.sensingresults.find({updatedAt : {$gt :"",$lte : "")}, presenceStatus : {$exists: 1}}).sort({updatedAt:1})
.forEach(function(doc) { 
    if (typeof prevSensingResults[doc.deviceId.toString()] !== undefined) {
         if (prevSensingResults[doc.deviceId.toString()].presenceStatus !== doc.presenceStatus && doc.presenceStatus === 0) {
             db.presenceagg.update({accountId: doc.accountId, buildingId: doc.buildingId, gatewayId: doc.gatewayId, deviceId: doc.deviceId, occupiedTime: prevSensingResults[doc.deviceId.toString()].updatedAt, vacantTime: doc.updatedAt}
         , {accountId: doc.accountId, buildingId: doc.buildingId, gatewayId: doc.gatewayId, deviceId: doc.deviceId, occupiedTime: prevSensingResults[doc.deviceId.toString()].updatedAt, vacantTime: doc.updatedAt, dwellPeriodInSeconds: (doc.updatedAt.getTime() - prevSensingResults[doc.deviceId.toString()].updatedAt.getTime()) / 1000}
                 , {upsert:true});
            prevSensingResults[doc.deviceId.toString()] = doc;
         } else if (prevSensingResults[doc.deviceId.toString()].presenceStatus !== doc.presenceStatus && doc.presenceStatus === 1) 
            prevSensingResults[doc.deviceId.toString()] = doc;
      }
    } else {
        prevSensingResults[doc.deviceId.toString()] = doc;
   }
})

Answer 2

Check if this solution meets your requirements.

Explanation

1. We join over the same collection. So for each item _i we take item _i+1. This method gives us where presenceStatus has been changed.
2. We filter document_{i i+1} pairs where presenceStatus is 0 - 1 or 1 - 0.
3. We group them into single data array.
4. Now we iterate over data by 2 steps (i=0;i<data.length;i+=2) and take updatedAt value.

    var occupiedTime = data[i].tmp.updatedAt
    var vacantTime   = data[i+1].tmp.updatedAt

5. We flatten calculated values and restore original document structure.

---

db.collection.aggregate([
  {
    $lookup: {
      from: "collection",
      let: {
        root_id: "$_id"
      },
      pipeline: [
        {
          $match: {
            $expr: {
              $gt: [
                "$_id",
                "$$root_id"
              ]
            }
          }
        },
        {
          $limit: 1
        }
      ],
      as: "tmp"
    }
  },
  {
    $match: {
      $or: [
        {
          "presenceStatus": 1,
          "tmp.presenceStatus": 0
        },
        {
          "presenceStatus": 0,
          "tmp.presenceStatus": 1
        }
      ]
    }
  },
  {
    $group: {
      _id: null,
      data: {
        $push: {
          $mergeObjects: [
            "$$ROOT",
            {
              tmp: {
                $arrayElemAt: [
                  "$tmp",
                  0
                ]
              }
            }
          ]
        }
      }
    }
  },
  {
    $addFields: {
      data: {
        $map: {
          input: {
            $range: [
              0,
              {
                $size: "$data"
              },
              2
            ]
          },
          as: "idx",
          in: {
            "occupiedTime": {
              $arrayElemAt: [
                "$data.tmp.updatedAt",
                {
                  $cond: [
                    {
                      $eq: [
                        {
                          $arrayElemAt: [
                            "$data.tmp.presenceStatus",
                            "$$idx"
                          ]
                        },
                        1
                      ]
                    },
                    "$$idx",
                    {
                      $add: [
                        "$$idx",
                        1
                      ]
                    }
                  ]
                }
              ]
            },
            "vacantTime": {
              $arrayElemAt: [
                "$data.tmp.updatedAt",
                {
                  $cond: [
                    {
                      $eq: [
                        {
                          $arrayElemAt: [
                            "$data.tmp.presenceStatus",
                            "$$idx"
                          ]
                        },
                        0
                      ]
                    },
                    "$$idx",
                    {
                      $add: [
                        "$$idx",
                        1
                      ]
                    }
                  ]
                }
              ]
            },
            "created": {
              $arrayElemAt: [
                "$data.tmp.createdAt",
                "$$idx"
              ]
            },
            "_id": {
              $arrayElemAt: [
                "$data.tmp._id",
                "$$idx"
              ]
            },
            "__v": 0
          }
        }
      }
    }
  },
  {
    $unwind: "$data"
  },
  {
    $replaceRoot: {
      newRoot: "$data"
    }
  },
  {
    $addFields: {
      "dwellTime": {
        $dateToString: {
          date: {
            $toDate: {
              $subtract: [
                "$vacantTime",
                "$occupiedTime"
              ]
            }
          },
          format: "%H-%M-%S"
        }
      }
    }
  }
])

MongoPlayground

Answer 3

You could use a Map-Reduce on the collection, this will output the difference in seconds, but that should be relatively easy to change on your end.

I haven't used this method on any sizeable collections, nor at all recently, so I'm not aware of performance implications by doing it in this fashion, please refer to the docs for more information.

db.collection.mapReduce(
  function() {
    emit(0, this);
  },
  function(key,values){
    var state = values[0].presenceStatus;
    var stateTimestamp = values[0].updatedAt;
    var result = {
      changes: []
    };

    for (var i = 1; i < values.length; i++){
      var value = values[i];
      if (value.presenceStatus !== state) {
        if (state === 1) {
          result.changes.push({
            "dwellTime": value.updatedAt - stateTimestamp,
            "occupiedTime": value.updatedAt,
          });
        }

        state = value.presenceStatus;
        stateTimestamp = value.updatedAt;
      }
    }

    return result;
  },
  {
    out: { inline: 1 }
  }
).results[0].value.changes;

[
        {
                "dwellTime" : 26979,
                "occupiedTime" : ISODate("2020-02-16T00:16:02.100Z")
        }
]