一种方法。不是最优雅但有效的例子：

list = LOAD 'shop' USING PigStorage(',') AS (category:chararray,id:chararray,mdate:chararray,my_store:chararray,item:chararray);
list_of = FOREACH list GENERATE category,SUBSTRING(mdate,0,4) AS my_date,my_store,item;
StoreG = GROUP list_of BY (category,my_date,my_store,item);
result = FOREACH StoreG GENERATE 
      group.category AS category,
      group.my_date AS my_date,
      group.my_store AS mys_store,
      group.item AS item, 
      COUNT(list_of.item) AS nb_items;
DUMP result;

当我们将别名项添加到 GROUP BY 语句基本上与查找不同的项然后对它们进行计数（正如您在括号中所做的那样）是相同的。
如果您仍然想使用您的代码，您只需添加一个关系 food_list.item 以下代码：

result = FOREACH StoreG
{
food_list = FOREACH list_of GENERATE item;
food_count = DISTINCT food_list;
GENERATE FLATTEN(group) AS (category,my_date,my_store),food_list.item,COUNT(food_count);
 }

StoreG = GROUP list_of BY (category,my_date,my_store);

应该是

StoreG = GROUP list_of BY (category,my_date,item);

因为您的预期结果是按项分组，而不是按存储区分组。