When a column in the CSV file contains mixed data types (e.g. numbers and strings) the ODBC driver uses the first few rows to detect the column's data type. If the first values are numbers, the ODBC driver tries to handle all the subsequent values as numbers as well and therefore the values that are strings (not numbers) will be ignored, resulting in a “null” value being scanned by mc.
A workaround for this behaviour is to prefix all values in the column with an arbitrary character, for example "#". The ODBC driver will then handle all values in this column as strings. Use a transformation rule in migration-center in order to remove the prefix "#" character from the values before they get assigned to the target attribute.