A Nonparametric Approach to Express the Influence of Chance On Validating Rules Discoveries

  • warning: Parameter 1 to theme_field() expected to be a reference, value given in /srv/www/vhosts/groups/kdmrg/includes/theme.inc on line 170.
  • warning: Parameter 1 to theme_field() expected to be a reference, value given in /srv/www/vhosts/groups/kdmrg/includes/theme.inc on line 170.
  • warning: Parameter 1 to theme_field() expected to be a reference, value given in /srv/www/vhosts/groups/kdmrg/includes/theme.inc on line 170.
The main goal of association rule mining (ARM) is to find all rules that associate dataset attributes to each other subject to minimum levels of support and confidence. However the occurrence of rules generated due to chance alone cannot be ruled out. Unrecorded and hidden factors that affect the variables in the datasets used in data mining increase the influence of chance on the discovered associations. In this talk I will demonstrate our investigation that we carried out to measure the influence of chance set-up when unsupervised association rule mining is used. I will also introduce a method for generating empirical random samples from an original dataset that preserve the characteristics of the original set when mining association rule. Our method also expresses the total effect of chance-setup on the generated results.