Discretization oriented to Decision Rules Generation

Many of the supervised learning algorithms only work with spaces of dis- crete attributes. Some of the methods proposed in the bibliography focus on the dis- cretization towards the generation of decision rules. This work provides a new dis- cretization algorithm called USD (Unparametrized Supervised Discretization), which transforms the infinite space of the values of the continuous attributes in a finite group of intervals with the purpose of using these intervals in the generation of decision rules, in such a way that these rules do not loose accuracy or goodness. Stands out the fact that, contrary to other methods, USD doesn't need parameterization.

[1]  Ron Kohavi,et al.  Supervised and Unsupervised Discretization of Continuous Features , 1995, ICML.

[2]  Jason Catlett,et al.  On Changing Continuous Attributes into Ordered Discrete Attributes , 1991, EWSL.

[3]  Chidanand Apté,et al.  Predicting Equity Returns from Securities Data , 1996, Advances in Knowledge Discovery and Data Mining.

[4]  Andrew K. C. Wong,et al.  Synthesizing Statistical Knowledge from Incomplete Mixed-Mode Data , 1987, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[5]  P E T R BERKA,et al.  empirical Comparison of Various Discretization Procedures , 1998, Int. J. Pattern Recognit. Artif. Intell..

[6]  Ron Kohavi,et al.  Error-Based and Entropy-Based Discretization of Continuous Features , 1996, KDD.

[7]  Huan Liu,et al.  Chi2: feature selection and discretization of numeric attributes , 1995, Proceedings of 7th IEEE International Conference on Tools with Artificial Intelligence.

[8]  Usama M. Fayyad,et al.  Multi-Interval Discretization of Continuous-Valued Attributes for Classification Learning , 1993, IJCAI.

[9]  Miguel Toro,et al.  Discovering hierarchical decision rules with evolutive algorithms in supervised learning , 2000, Int. J. Comput. Syst. Signals.