Abstract
The paper describes application of machine learning in real physical domain - identification of subatomic particles on the basis of data obtained from electromagnetic calorimeter. Two machine learning systems, MAGNUS ASSISTANT and RETIS were used to generate decision tree which can identify electron and pion using 18 attributes obtained from the calorimeter measurements. During the analysis it was found out that most of the attributes can be omitted without seriously affecting the classification accuracy. Using less attributes yields smaller and more understandable decision trees, which are preferred by the experts.