B018 Cluster analysis for log data for thin Miocene beds. Summary 1 An accurate lithological prediction using wireline logs in the Miocene sediments in the Carpathian Foredeep is not a simple problem due to the thin interbedded sandy-shale sequences (Fig. 1.) [Dziadzio 2000; Karnkowski 1999]. Statistical methods i.e. principal component analysis (PCA) and discriminant analysis (DA) were applied for selecting a combination of the most informative logs for determining in detail sandstone beds and reducing number of data [Kazmierczuk-Kotula 2004]. Because we were not fully satisfied with the results of previous approach other statistical method i.e. cluster analysis was used