问题描述
我想通过减少COUNT()操作的数量来减少处理量。 例如,如果conditionA是字段不是NULL,则conditionB是字段是NULL。 我想对数据中的项目总数进行计数,并用conditionA减去项目数,以便获得conditionB的计数。 本质上,数学方程就是conditionB = total-conditionA。
filterByTime = FILTER data BY timestamp >= 0;
filterByTimeGroup = GROUP filterByTime ALL;
filterByTimeCount = FOREACH filterByTimeGroup GENERATE COUNT(filterByTime);
filterByConditionA = FILTER filterByTime BY conditionA IS NOT NULL;
filterByConditionAGroup = GROUP filterByConditionA ALL;
filterByConditionACount = FOREACH filterByConditionAGroup GENERATE COUNT(filterByConditionA);
filterByConditionB = filterByTimeCount - filterByConditionACount;
我将如何在Apache Pig中做到这一点?我是否需要使用SPLIT运算符并从这些组中进行计数?
解决方法
计数,其中conditionA为null。请注意,这是假定没有基于其他过滤器的其他子集。
filterByConditionB = FILTER data BY (conditionA IS NULL AND timestamp >= 0);
filterByConditionBGroup = GROUP filterByConditionB ALL;
filterByConditionNCount = FOREACH filterByConditionBGroup GENERATE COUNT(filterByConditionB);