The named characteristics are quite clear. Of 16,085 order documents overall, only 15,882 are actual order documents, the remainder splitting into 73 contracts, 122 scheduling agreements, and 8 requests for quotations.
To avoid errors in data analysis (e.g. it would be very misleading to present values for contracts and orders and requests undifferentiated in a single total) it makes sense to differentiate analysis by document type, or to exclude those that are not relevant right from the start.
2. Do not ignore what is deleted
There are many examples in which data records deleted in data analysis, or data records identified as deleted can be ignored. Proper deletion occurs in the rarest cases because of the obligation to retain or record. Instead, data (both master data and transactions) exhibit an attribute often called a "delete flag". If an "X" is entered for instance, the document or master data are marked as deleted. This is how the term "deleted" is used in this article too. If I write "deleted" this means data records with an appropriate delete flag, but that may easily still be found in the tables. This is less likely to occur in financial transactions because there is an offsetting entry or cancellation. More frequently however in advance processes like purchasing or sales, or the associated master data. In the latter case it can sometimes really make sense to shut out data records marked as deleted: If you analyze duplicates in vendor master records, for example, to avoid double payment, or enhance data quality, deleted master records should be excluded from analysis because these cannot be used per se in bookings. A division would not react positively if the list with potential vendor duplicates also contained master records ready labeled for deletion.
The situation is different with purchasing documents. Here you find a possible delete flag both at header level (table EKKO) and at item level (table EKPO). As a rule, single items can be marked as deleted, but the whole document can only be labeled for deletion if all items in it are deleted.
In the purchasing area however, it should be noted that deleted orders or order lines should not absolutely be excluded from analysis. The reason is that often, from the viewpoint of ordering operation, orders or order lines are marked as deleted if they are seen to be finished. That should prevent their further use, although the activity is finished in content (even if there are still quantities remaining in an order that were not yet called for and released).
But that in turn means that there may quite easily be goods receipts and invoice receipts for an order item that are interesting for analysis. If you excluded orders or order items marked as deleted, you would entirely miss out these operations, and the resulting picture of quantity and value structure of ordering operations would be a false one.