An Efficient and Scalable Algorithm for Mining Maximal High Confidence Rules from Microarray Dataset

DNA microarrays allow simultaneous measurements of expression levels for a large number of genes within a number of different experimental samples. Mining association rules algorithms are used to reveal biologically relevant associations between different genes under different experimental samples. In this paper, we present a new mining association rules algorithm called Mining Maximal High Confidence Rules (MMHCR). The MMHCR algorithm is based on a column (gene) enumeration method which overcomes both the computational time and memory explosion problems of column- enumeration method used in many of the mining microarray algorithms. MMHCR uses an efficient data structure tree in which each node holds a gene's name and its binary representation. The binary representation is beneficial in two folds. First, it makes MMHCR easily find all maximal high confidence rules. Second, it makes MMHCR more scalable than comparatives. In our experiments on a real microarray dataset, MMHCR attained very promising results and outperformed other counterparts.