Data mining systems suitable for data mining teaching have been investigated in the paper. Ussualy, such systems as SPSS Modeler (Clementine), Statistica, SAS/STAT are used in the mathematical statistics courses. However, they are not always suitable for data mining teaching. WEKA, Orange, KNIME, RapidMiner systems are more appropriate for this purpose.

The aim of the research is to prepare proposals for assurance of the business clusters formation regarding the regional development potential upon study of the clusterization of Lithuanian regional companies. After extensive analysis of scientific literature, the qualification of the cluster, their structure, main features, goals and benefits to its members, region, and sector where cluster is based, are observed. From analysis of relevant secondary data the main problems that hinder the development of clusters in distinct regions of Lithuania are distinguished. It is revealed that clusterization in Lithuanian regions lags far behind big cities, it is much more passive and clusters there often reach only the level of a micro-cluster. Clusters are most developed in the tourism and food industries, using traditional means instead of high-techs, failing to benefit from EU support for clustering development. In order to improve the clusterization situation in Lithuania and its development in the regions, more attention and investment are to be allocated for promoting cooperation between the companies and the business and science, research sectors, and joining the international cluster. The state support should be prioritized in the rural regions locating less clusters, forming a reliable means and communication network for these clusters’ development.

Rational models of electoral behavior emphasize the need of sufficient information for voters to make their decisions. Monitoring the behavior of a single politician is not easy to implement, not to mention of the whole parliament, since for the latter one must apply statistical methods designed for the analysis of large amounts of information. In this paper we propose methods and techniques for the analysis of voting behavior of the Lithuanian Parliament (Seimas) that allow for clearer identification and recognition of voting patterns of the Seimas. Votes of the last sessions of the 2008-2012 term of the Seimas (pre-election period) are analyzed employing cluster analysis. Also, multidimensional scaling is used to visualize the generated results. Results obtained using different vote coding methods and clustering techniques are compared in the paper, too.

Racionalūs rinkiminės elgsenos modeliai akcentuoja pakankamos informacijos poreikį rinkėjams priimant sprendimus, už ką balsuoti. Stebėti net pavienių politikų elgseną nėra paprasta, o bandyti fiksuoti ir suprasti viso parlamento narių veiklą yra dar sudėtingesnis uždavinys, nes dideliam kiekiui informacijos apdoroti būtina taikyti statistinius metodus. Šiame darbe siekiama pasiūlyti tinkamas metodikas LR Seimo balsavimų stebėsenai, leidžiančias aiškiau identifikuoti parlamentarų balsavimo tendencijas. Statistiškai analizuojami LR Seimo balsavimai 2008-2012 metų kadencijos pabaigoje (priešrinkiminiu laikotarpiu). Naudojamos klasterizavimo procedūros, o gauti rezultatai daugiamačių skalių metodo pagalba vaizdžiai pateikiami grafiškai. Lyginami skirtingi balsavimų kodavimo ir klasterizavimo metodai.