I am a newbie for this area of research, and just started getting my hands wet testing some ideas.
One problem I have is my data has a column that is a categorical input selected from about 1,000 string values. I could use one-hot vector to encode the values, but because each of the category represents a group of elements, and these categories have group-subgroup relationships, it would be much favorable to have this kind of relationship encoded for learning. I could not find answers online. I would appreciate it very much if you guys could provide some insightful thoughts. Many thanks!!