The article presents a tool to analyze the application of efficient algorithms of data mining, namely hierarchical clustering algorithms to be used in the analysis of geological data. It introduces a description of hierarchical clustering principles and methods for learning dependencies from geological data. The authors are using statistical formulation of algorithms to represent the most natural framework for learning from data. The geological data come from mining holes, and describe the structure of sedimental layers of vertical section of geological body. The analysis of such data is intended to give a basis for uniform description of lithological characteristics, and for the identification of them via formal methods.