Given a non-normalized data table, is there a good a strategy to attempt to normalize based on repeated data? Assume we have no information about the significance of the columns.
In a small example
Col1 Col2 Col3 ======================================== Europe France Paris Europe France Marseilles Europe England London Europe England Manchester N.America USA St. Louis N.America USA Washington D.C.
I can see that there are three source tables, probably called,
Cities. What about for much larger data sets. Say 45 columns and several hundred thousand rows. Looking at the data is obviously not going to help much.