group,aggregate,mapReduce 分组统计: group() 简单聚合: aggregate() 强大统计: mapReduce() db.collection.group(document) document:{ key:{key1:1,key2:1}, //根据那几个字段分组 cond:{}, //筛选的条件 reduce: function(curr,result) { //分组之后的聚合运算,curr是一行数据,result是计算后的结果 }, initial:{}, //初始化result里面 finalize:function() { //reduce一组都执行完毕后最后执行的函数 } } #计算每个栏目下(cat_id)的商品数 count()操作 select cat_id,count(*) from goods group by cat_id; //mysql操作 use shop db.goods.group( { key:{cat_id:1}, //根据哪个字段分组 cond:{}, //所有行取出来,不加条件 reduce:function(curr,result) {//reduce的执行过程:每一行就是一个curr,每一组共用一个result变量, result.cnt += 1; //result.cnt是每组有多少行,每个组有一个result, }, initial:{cnt:0} } ): [ { "cat_id" : 4.0, "cnt" : 3.0 }, { "cat_id" : 8.0, "cnt" : 3.0 }, { "cat_id" : null, "cnt" : 2.0 } ] #查询每个栏目下价格高于3500元的商品数量 use shop db.goods.group( { key:{cat_id:1}, //cat_id分组,并且查出car_id和shop_price字段 cond:{shop_price:{$gt:3500}}, reduce:function(curr,result) { result.cnt += 1; }, initial:{cnt:0} } ): [ { "cat_id" : 3.0, "shop_price" : 5999.0, "cnt" : 1.0 }, { "cat_id" : 5.0, "shop_price" : 3700.0, "cnt" : 1.0 } ] #查询每个栏目下价格大于3000元的商品个数 { key:{cat_id:1}, cond:{}, reduce: function(curr,result) { result.total += 1; }, initial:{total:0} }: [ { "cat_id" : 4.0, "total" : 3.0 }, { "cat_id" : 8.0, "total" : 3.0 }, { "cat_id" : null, "total" : 2.0 } ] #计算每个栏目下的商品库存量 sum()操作 select sum(goods_number) from goods group by cat_id; use shop db.goods.group( { key:{cat_id:1}, cond:{}, reduce: function(curr,result) { result.total += curr.goods_number; }, initial:{total:0} } ): [ { "cat_id" : 4.0, "total" : 3.0 }, { "cat_id" : 8.0, "total" : 61.0 }, { "cat_id" : null, "total" : NaN } ] #查询每个栏目最贵的商品价格, max()操作 select max(shop_price) from goods group by cat_id; use shop db.goods.group( { key:{cat_id:1}, cond:{}, reduce:function(curr , result) { if(curr.shop_price > result.max) { result.max = curr.shop_price; } }, initial:{max:0} } ): #查询每个栏目下商品的平均价格 select cat_id,avg(shop_price) from goods group by cat_id; use shop db.goods.group( { key:{cat_id:1}, //相当于group by cond:{}, //相当于where reduce:function(curr , result) { //相当于sum.avg函数 result.cnt += 1; result.sum += curr.shop_price; }, initial:{sum:0,cnt:0}, //进这个组执行一下 finalize:function(result) { //出这个组执行一下, 组操作完毕后的回调函数 result.avg = result.sum/result.cnt; } } ): [ { "cat_id" : 4.0, "sum" : 6891.0, "cnt" : 3.0, "avg" : 2297.0 }, { "cat_id" : 8.0, "sum" : 226.0, "cnt" : 3.0, "avg" : 75.3333333333333 }, { "cat_id" : null, "sum" : NaN, "cnt" : 2.0, "avg" : NaN } ] 注意: 1:group需要我们手写聚合函数的业务逻辑 2:group 不支持集群shard cluster, 无法分布式运算 3:分布式可以用 aggregate() (version2.2) , 或者mapReduce() (version2.4) GROUP BY $group HAVING $match SELECT $project ORDER BY $sort LIMIT $limit SUM() $sum COUNT() $sum #查询每个栏目下的商品数量 select count(*) from goods group by cat_id; db.goods.aggregate( [ { $group:{ _id:"$cat_id", //根据cad_id分组 total:{$sum:1} //乘以1 } } ] ): { "_id" : null, "total" : -2.0 } { "_id" : 14.0, "total" : -2.0 } { "_id" : 2.0, "total" : -1.0 } { "_id" : 13.0, "total" : -2.0 } #查询goods下有多少条商品,select count(*) from goods [ {$group:{_id:null,total:{$sum:1}}} ]; { "_id" : null, "total" : 33.0 } #查询每个栏目下 价格大于3000元的商品个数 use shop db.goods.aggregate( [ {$match:{shop_price:{$gt:3000}}}, {$group:{_id:"$cat_id",total:{$sum:1}}} ] ): { "_id" : 5.0, "total" : 1.0 } { "_id" : 3.0, "total" : 2.0 } #查询每个栏目下 价格大于50元的商品个数 #并筛选出"满足条件的商品个数" 大于等于3的栏目 select cat_id,count(*) as cnt from goods where shop_price>3000 group by cat_id having cnt>=2 use shop db.goods.aggregate( [ {$match:{shop_price:{$gt:3000}}}, //放在group之前是where {$group:{_id:"$cat_id",total:{$sum:1}}}, {$match:{total:{$gte:2}}} //放在group之后是having ] ): { "_id" : 3.0, "total" : 2.0 } #查询每个栏目下的库存量 use shop db.goods.aggregate( [ {$group:{_id:"$cat_id" , total:{$sum:"$goods_number"}}}, //cat_id分组,goods_number求和, ] ): { "_id" : 5.0, "total" : 8.0 } { "_id" : 15.0, "total" : 2.0 } #查询每个栏目下的库存量,并按库存量排序 use shop db.goods.aggregate( [ {$group:{_id:"$cat_id" , total:{$sum:"$goods_number"}}}, {$sort:{total:1}} //1是升序 ] ) #查询每个栏目下的库存量,并按库存量排序 use shop db.goods.aggregate( [ {$group:{_id:"$cat_id" , total:{$sum:"$goods_number"}}}, {$sort:{total:1}}, {$limit:3} //取前3个 ] ): { "_id" : null, "total" : 0 } { "_id" : 2.0, "total" : 0.0 } { "_id" : 15.0, "total" : 2.0 } #查询每个栏目的商品平均价格,并按平均价格由高到低排序 select cat_id ,avg(shop_price) as pj from goods group by cat_id order by pj desc limit 3 use shop db.goods.aggregate( [ {$group:{_id:"$cat_id" , avg:{$avg:"$shop_price"}}}, //car_id排序,shop_price求平均, {$sort:{avg:-1}}, {$limit:3} ] ): { "_id" : 5.0, "avg" : 3700.0 } { "_id" : 4.0, "avg" : 2297.0 } { "_id" : 3.0, "avg" : 1746.06666666667 }
mapReduce 随着"大数据"概念而流行,mapReduce的真正强项在于分布式。 其实mapReduce的概念非常简单,比aggregate要简单,从功能上说,相当于RDBMS(传统数据库)的 group 操作。 当数据非常大时,像google,有N多数据中心,数据都不在地球的一端,用group力所不及.group既然不支持分布式, 由于单台服务器的运算能力必然是有限的. 而mapRecuce支持分布式(不是算法好),而是支持大量的服务器同时工作,用蛮力来统计.mapRecuce就是group和aggregate,只不过支持分布式。 mapRecuce的工作过程:1.map-->映射,2.reduce->归约 map: 1.先在全世界机器找(分布式集群上找),把属于同一个组的数据,映射到一个数组上.cat_id [23,2,6,7],2.reduce: 把数组(同一组)的数据,进行运算. #用mapReduce计算每个栏目的库存总量 //map函数(进行映射工作,映射成一个二维数组) var map = function() { emit(this.cat_id,this.goods_number); //根据cat_id分组, } /* { cat_id1:[goods_number1,goods_number2,goods_number3.....], cat_id2:[goods_number1,goods_number2,goods_number3.....] cat_id3:[goods_number1,goods_number2,goods_number3.....] } */ var reduce = function(cat_id,numbers) { //对数组做处理,求goods_number的和, return Array.sum(numbers); //mongo对js的数组增加的求和方法 } /* { _id:cat_id1, value:goods_number1+goods_number2+goods_number3....., _id:cat_id1, value:goods_number1+goods_number2+goods_number3....., _id:cat_id1, value:goods_number1+goods_number2+goods_number3....., } */ db.goods.mapReduce(map,reduce,{out:'res'}); //out计算的结果放在res集合里面去, //多了一个res表 show tables db.res.find(): { "_id" : null, "value" : NaN } { "_id" : 2.0, "value" : 0.0 } { "_id" : 3.0, "value" : 203.0 } { "_id" : 4.0, "value" : 3.0 } { "_id" : 15.0, "value" : 2.0 } //查看array的所有方法: for (var k in Array){ print(k) }: contains unique shuffle tojson fetchRefs sum avg stdDev #用mapReduce计算每个栏目下商品的平均价格 var map = function() { emit(this.cat_id,this.shop_price); } var reduce = function(cat_id,values) { return Array.avg(values); } db.goods.mapReduce(map,reduce,{out:'res'}); : { "_id" : null, "value" : NaN } { "_id" : 2.0, "value" : 823.33 } { "_id" : 3.0, "value" : 1746.06666666667 }
var map = function() { if(this.jing < 0 || this.wei < 0){ return; } var j = Math.floor(this.jing/5)*5; var w = Math.floor(this.wei/5)*5; var block = j+":"+w; emit(block,1); } var reduce = function(block,values) { return Array.sum(values); } db.goods.mapReduce(map,reduce,{out:'res'});