P000012

*Po-wei Hua (National Changhua University of Education (NCUE))

One the one hand, in the system of datamining, ''cluster'' is a group of the same elements gathered; instead, it makes more differences between groups. Thus, it makes us easier to design something suitable for each group and help us to reduce the amount of data-using. This system has been developed for years. It includes many kinds of algorithm, such as k-mean. While on the other hand, ''heterogeneous cluster'' is just the opposite of cluster. It is a group of differences elements gathered; instead, it makes more similar between groups. Nowadays, seldom research about heterogeneous cluster has been published. However, it was widely-used in our daily life, such as distribution of classes, balanced diet, group sampling, etc. Nowadays, the application of heterogeneous cluster is in the field of education, including Random cluster and S cluster. Random cluster is not ideal enough, for the students are distributed randomly without any specific heterogeneous target. As for S cluster, it uses heterogeneous as the target to cluster the group. However, after analyzing the simulated data, the higher the dimension is, the more similar the result of Random cluster and S cluster is. Therefore, S cluster turns out to be a unsuitable method either. In fact, it is impossible to find a ''perfect'' solution. Namely, it's a NP- hard question. Therefore, I created another heterogeneous cluster algorithm system to find a second best solution which is similar to Ripple. It diffuses from the core so I named it as Ripple Cluster. After analyzing the simulated data by using the Ripple Cluster, the result has a great chance to be more ideal than Random cluster and S cluster.

Math formula preview: