Spark sql HLL函数

HLL函数：

预估函数：presto_estimate_prepare（预估计算）

合并函数：presto_estimate_merge（创建binary格式数据，用于后续计算）

计算函数：presto_estimate_compute（对binary格式数据进行计算，类似拟合计算count distinct）

概念说明：https://www.bookstack/read/Doris/spilt.9.spilt.7.11.md

 --presto_estimate_compute(`hll_arry列`)
SELECT 
	t_a.scene_of_flow_name                            						AS `活动类型` 
	,t_b.new_category_1st_name  										AS `一级类目` 
	,t_b.new_category_3rd_name                            			  	AS `三级类目` 
	,sum(pv_1d)                                                                      AS`曝光页面数`
	,presto_estimate_compute(uv_hll_1d) 							AS `UV` 
	,sum(goods_acture_amt_1d)											AS `销售额(含退拒)` 
	,sum(result_goods_acture_amt_1d)								AS `销售额(不含退拒)` 
	,sum(goods_acture_num_1d)										AS `销量(含退拒)` 
	,sum(result_goods_acture_num_1d )							AS `销量(不含退拒)` 
	,sum(cart_cnt_1d)                                                            AS `加购数` 
	,presto_estimate_compute(user_num_hll_1d) 			AS `客户数` 
	,presto_estimate_compute(order_num_hll_1d) 			AS `订单数` 
	,count(DISTINCT t_b.spu_id)                                          AS `SPU数` 
	,count(DISTINCT t_a.merchandise_no)                        AS `商品数` 
FROM
	t_a
GROUP BY
	t_a.scene_of_flow_name