Mining Frequent Item Sets Efficiently by Using Compression Techniques

Mining frequent item sets in a data set is a significant problem of data mining that can only be solved in exponential time. For especially very large data sets, finding frequent item sets in practical run times is extremel y important. A market basket data set can be represented by a set of bit vectors, which enables fast computation and low memory requirements. Space requirements can be reduced and better run time results can be obtained if bit vectors can be compressed, and binary operations can be applied on compressed bit vectors. In this paper, we study the advantages and disadvantages of using compression techniques fo r finding frequent item sets. Experimental evaluations clear ly show that applying special purpose compression techniques has many benefits on a wide range of data sets.