Some data are messy and column can be a mix of different type of data. For example if you work on an address database, you might have a ZIP code column and a State column. And some ZIP code can be stored in the State Column.
1. Using the facet functionality of refine, identified all ZIP code stored in the State column. A quick way to do this is to facet by Text length facet and select all entries that have more than 2 letters (hopefully, your database contains only the 2 ID for the state and not the full name).
You can use the text facet to check the data quickly.
2. Select the ZIP column and invoke edit cells => transform.
3. enter the following expression:
cells["State"].value
And you're done.
Be careful, entries in the State column will be erased. Save (move) them to an other column if you want to keep track of them.