Mining dynamic interdimension association rules for local-scale weather prediction

Abstract

Mining dynamic interdimension association rules for local-scale weather prediction is to discover abnormal weather phenomena changing so that the professional weather forecaster can use these rules to predict some severe weather situations, such as hail storm, thunder storm and so on. A weather analysis is composed of individual analyses of the several meteorological variables. When some of meteorological variables have some special change tendency, some kind of severe weather will happen in most cases. We propose a new algorithm, DIAL to discover potential relations between the special change tendency and the severe weather. The algorithm consists three parts: (1) Change the original static database recording the weather condition data into a new database with the changing tendency of every measurements of the weather; (2) Discover multidimensional association rules from the new generated database; (3) Use the predefined predicts to transfer the interval rules into the dynamic interdimension association rules.