按记录计数将子记录划分为簇，而不拆分父ID

I'm trying to build a process to split a group of transactions into roughly equal sized groups to facilitate chunking the data.I'm trying to build a process to split a group