AUTOMATIC DISCRETIZATION IN PREPROCESSING FOR DATA ANALYSIS IN MOBILE NETWORK

Pekka Kumpulainen, Kimmo Hätönen, Pekko Vehviläinen
Abstract:
Several data mining methods require discretization of the data, rough sets and association rules for example. In this paper we present simple discretization methods for two types of variables common in telecommunications network. The variables are first classified in order to apply the proper discretization method. These methods are based on the amplitude distribution of the variables. Methods are unsupervised, i.e. they do not require class information for the samples.
Keywords:
preprocessing, discretization, telecommunications network
Download:
PWC-2003-TC4-083.pdf
DOI:
-
Event details
Event name:
XVII IMEKO World Congress
Title:

Metrology in the 3rd Millennium

Place:
Dubrovnik, CROATIA
Time:
22 June 2003 - 28 June 2003