Split


Synopsis

Creates new attributes from a nominal attribute by dividing the nominal values into parts according to a split criterion.


Description

This operator creates new attributes from a nominal attribute by dividing the nominal values into parts according to a split criterion (regular expression). This operator provides two different modes, depending on the setting of the parameter "splitting_mode".

Ordered Splits

In the first split mode, called ordered_split, the resulting attributes get the name of the original attribute together with a number indicating the order. For example, if the original data contained the values attribute-name


value1 value2, value3 value3

and should be divided by the separating commas, the resulting attributes would be attribute-name1, attribute-name2, attribute-name3 with the tuples (value1, ?, ?), (value2, value3, ?), and (value3, ?, ?), respectively. This mode is useful if the original values indicated some order like, for example, a preference.

Unordered Splits

In the second split mode, called unordered_split, the resulting attributes get the name of the original attribute together with the value for each of the occurring values. For example, if the original data contained the values attribute-name


value1 value2, value3 value3

and again should be divided by the separating commas, the resulting attributes would be attribute-name-value1, attribute-name-value2, and attribute-name-value3 with the tuples (true, false, false), (false, true, true), and (false, false, true), respectively. This mode is useful if the order is not important but the goal is a basket like data set containing all occurring values.


Input


Output


Parameters


ExampleProcess