An Application of Inductive Learning

The development of bar-code technology provided accurate and large market databases for researchers who deal with datasets. Since the data is large both in dimension and size, most exploratory analysis techniques of statistics are not appropriate for such tasks. In this paper, we describe a high-level algorithm, and the application of it on a large basket data, extracted from the database of a big supermarket company. The algorithm have two consecutive steps. Each step is a diierent popular machine learning method: clustering and classiication. In this application, we used KMEANS clustering algorithm and C4.5 classiication program respectively. At the end of the application we come up with a set of items that can be employed for promotion. By promotion we aim to increase number of costumers that make their weekly or monthly shopping, which refer to full baskets among transactions.