Knowledge Discovery of Interesting Classification Rules Based on Adaptive Genetic Algorithm
Yong Zhou · 2007
Data classification is a very important point in DataMining, but the existing classification algorithms always only discover the classification rules with high accuracy, and the research about interesting classification rules is few.So this paper proposes an algorithm to find the interesting classification rules based on Genetic Algorithm.Firstly, we design the fitness function with the attributes' information gain, and the settings of weights of the information gain, and the interestingness of the rules, so we combine the objective and subjective measure methods together.Secondly, we use the adaptive genetic algorithm to keep the process from constringency early, and then we can reduce the convergence speed.At last, the results of the experiment given by JBuilder2006 can discover the interesting classification rules, illustrating the effectiveness of this algorithm.