The speed of a drilling process is predominantly influenced by the properties of the penetrated formation. Drilling plans are mostly based on previous experience in the field (if exists) and the expected geological circumstances. Although geological information is rarely 100% accurate, non-expected geological formations could still cause trouble, thus a tool which could provide information about current geological circumstances could be in great use. Studies show that there is a connection between drilling parameters and formation properties. In this thesis work, an investigation will take place on the correlation between drilling- and geological data of existing wells, located in Austria. Statistical methods will be implemented on depth- (and time-)based drilling data to create well defined groups which can be connected to a certain type of formation. The preparation of input data set for the machine learning algorithm is a crucial task of such a work. Different types of filtering, harmonization of additional data sets, and various plotting techniques will be presented in the thesis. The scope of this work is to find the connection between drilling and formation parameters and then develop an algorithm which can determine the geological formation based on drilling data, using machine learning.