How to Prepare an Input Data File (part 2)
N - number of examples .............. max. 1000
M - number of input attributes ....... max. 1000
W - number of characters in an attribute name or value .... max. 30
Process of input data preparation begins with its first row. First row should contain M+1 attribute names. Examples of valid attribute names for DMS are:
Attribute names must always be separated by a delimiter. Attribute names must begin with a letter (a-z or A-Z) or an underscore (ASCII 95 decimal) but it is a good practice to start with an upper-case letter (A-Z).
Except letters, attribute names may include digits (0-9) and underscores. Spaces (ASCII 32 decimal) are allowed if space is not the delimiter. Maximal attribute name length including spaces is W characters.
All attribute names specified in the first row of a data file must be different.
Target attribute name is an attribute name which has exclamation mark (!) before its first character. Position of the target attribute name in the first row is not fixed. Example of a good target attribute name is:
If the user wants to exclude some attribute from the rule induction process she/he should put a question mark instead of an attribute or in front of an attribute name. Examples of excluded attribute names with '?' are:
Maximal number of input attributes is 1000, regardless if their names begin with '?' or not.
© 2001 LIS - Rudjer Boskovic Institute
Last modified: September 09 2015 14:17:42.