Seminář: Relational datamining extensions for 4FT and SD4FT in Ferda

Existing datamining methods are based upon single-table mining. Although this restriction simplifies the approach to implementation, it could be limiting in practical applications, where the data is represented by several tables in relation. Relational datamining broadens the classical approach and enables working with several datatables in relation. It defines the concept of virtual attribute that adds virtual columns to the main data table as a result of either simple data transformation or the result of datamining subtask run on the table in relation. The presentation will describe the implementation of relational extensions for 4FT and SD4FT procedures for the environment Ferda DataMiner. The general approach will be shown and its advantages and drawbacks will be discussed. Some practical examples will be shown as well. After that, possibilities of further development will be briefly presented.