So in fact the data set cannot be split exactly in two. Perhaps you should restate your "requirement".
And any/all rules for processing.
Do understand that the biggest part of getting help from a forum is clearly/com;pletely describing what you want to do.
The best way to explain something like this is to post (in paragraph form) your complete requirement - including some sample input data and the exact output you want from that input data. You do not need high volume. You do not need full-width records. Just enough to demonstrate the possible input and the needed output.
When posting data, you should use the Code tag to preserve alignment and improve readability. There is a Preview function so you can see your post as it will appear to the forum rather than how it appears in the Reply editor.