使用Pig on Hadoop通过层分配活动频率的用户

Question

需要计算每个组中的用户数量，并且每个用户的组分配基于他们的付款（例如，付款少于2，或介于3到5之间，或超过5）。这是我目前的代码，想知道是否有任何方法可以使它更优雅？是否可以通过一组语句完成逻辑？感谢。

customer_group = group payments_feed by customerID;
customer_payment_count=foreach customer_group generate customerID, COUNT(payments_feed) as payment_amount;
tier1 = filter customer_payment_count by payment_amount <= 2;
tier2 = filter customer_payment_count by 3 <= payment_amount <= 5;
tier3 = filter customer_payment_count by payment_amount > 5;
tier1_group = group tier1 by all;
tier1_count = foreach tier1_group generate COUNT_STAR(tier1);
tier2_group = group tier2 by all;
tier2_count = foreach tier2_group generate COUNT_STAR(tier2);
tier3_group = group tier3 by all;
tier3_count = foreach tier3_group generate COUNT_STAR(tier3);
result = UNION tier1_count, tier2_count, tier3_count;

一些虚假数据，架构客户ID（唯一），付款（价值始终为1，因为它代表客户付款的时间），

1 1
2 1
1 1
3 1
4 1
1 1
2 1
1 1
1 1
5 1

在这种情况下，客户1进行了5次付款，应该是第2级，而所有其他客户付款不超过2次，它们都属于第1层。

所以，期望的输出是，

4 1 0

提前谢谢， 林

Answer 1

没有。 GROUP 足以计算相同的内容。因为要计算特定包中的元组数，所以使用了额外的 GROUP 。但不是使用三个 FILTER 。您可以使用 SPLIT ，代码如下：

customer_group = group payments_feed by customerID;
customer_payment_count=foreach customer_group generate customerID, COUNT(payments_feed) as payment_amount;
split into customer_payment_count into tier1 if(coun<=2), tier2 if(coun>=3 AND coun<=5),tier3 if(coun>5);
tier1_group = group tier1 by all;
tier1_count = foreach tier1_group generate COUNT_STAR(tier1);
tier2_group = group tier2 by all;
tier2_count = foreach tier2_group generate COUNT_STAR(tier2);
tier3_group = group tier3 by all;
tier3_count = foreach tier3_group generate COUNT_STAR(tier3);
result = UNION tier1_count, tier2_count, tier3_count;