If I understand it correctly in DataAnalysis the wildmatch function is operating on a longer string than in DataRaw. So it may encounter more matches in the part of the string that is not processed in DataRaw because of the mid(@1, 22).
You have one DataRaw table and two loads the load data in that table. The result of both loads are concatenated because they have exactly the same fields so the result is one table with records from both loads. In one 'DataRaw' load you do a wildmatch and that processes only the records you then load into the DataRaw. But in DataAnalysis you process the data that was loaded into the DataRaw table with both loads. So the wildmatch in DataAnalysis processes more rows.