Merging a lookup table
Nominal variables with more than several categories pose a potential problem. First, fields with a large number of categories can significantly increase processing time. Second, these fields can potentially have categories with very few cases, which can become problematic (for example, they might be outliers or just difficult to understand). Third, these fields might not even be used by certain models (see the following screenshot). Finally, fields with a large number of categories might not really get at the crux of the real characteristics of interest. Many new users of Modeler don't realize that many algorithms are automatically transforming nominal variables behind the scenes. Within the General Setting in Stream Properties, there are two options designed to prevent this problem from getting out of hand.
As mentioned earlier, many times fields with a large number of categories might not really get at the real characteristics of interest and therefore sometimes...
