markdown PostPogo聚合通过timebuckets -

Posted 2021-05-06

To aggregate timeseries based off of post types, need two values - 

1. the timebucket wrapper
2. the type of time series you want (days, minutes, weeks, etc...)

Below gives me aggregation for Feb. 2015 by type by minute -

    db.post.aggregate( 
      {$match: {timebucket: "2015-1-m"}}, 
      {$unwind: "$timebucket"},    
      {$match: {timebucket:{$regex : ".*-min$"}}},    
      {$project: {tb_type: {$concat: ["$timebucket", "/", "$type"]}}},
      {$group: {_id: "$tb_type", count: {"$sum": 1}}} );
    

Results - 

    { "_id" : "2015-1-11-21-1-min/vine", "count" : 1 }
    { "_id" : "2015-1-11-12-1-min/vine", "count" : 1 }
    { "_id" : "2015-1-7-0-1-min/vine", "count" : 1 }
    { "_id" : "2015-1-4-23-1-min/vine", "count" : 1 }
    { "_id" : "2015-1-1-16-1-min/vine", "count" : 1 }
    { "_id" : "2015-1-1-14-1-min/vine", "count" : 1 }

Example Timebuckets - 

    [ 
      '2014-7-1-10-7-min',
      '2014-7-1-10-h',
      '2014-7-1-d',
      '2014-7-5-dowm',
      '2014-5-dowy',
      '2014-7-m',
      '2014-y',
      '2014-31-w',
      '31-wa',
      '5-dowa' 
    ]