Effectiveness of ordinal information for data mining

During the last decade the technologies for generating and collecting data have advanced rapidly. As a result, the problem now is not the obtaining of data but the techniques, able to analyze large volumes of data and to produce meaningful and useful information. One of the most popular data mining tasks is that of classification. Though data mining is oriented to analyzing large volumes of data of different nature (quantitative as well as qualitative ones), additional knowledge about dependencies among all elements of these data sets may change the results of the analysis from failure to success. In some classification tasks classes and attribute values are connected in an ordinal way. We show that if we take into account ordinal dependencies among data elements, we may produce much more manageable and meaningful results