There seems to have been a misunderstanding here.
When importing data from Excel, you can define how to import that data. You may/must specify which data defines the graph structure, which data is imported into custom properties, and which data is simply ignored.
In any case, after yEd has finished the import, any ignored data or data yEd could not handle is discarded and cannot be reproduced in any way when saving to GraphML format. But, of course, all data that *was* imported (including custom properties) will be stored in GraphML. In other words, any information that is available in yEd *after* the import is also available in GraphML.