IFRS 9 a modelování očekávaných kreditních ztrát

The key aim of this thesis is to compare two approaches of modelling expected credit loss, as are required by the accounting standard IFRS 9. The theoretical part of the thesis focuses on history of IFRS 9, its introduction and description of components entering the ECL calculation in the form as required by the standard. In the practical part implementation of two of the approaches is performed on data from P2P lending platform Bondora. The two approaches are using survival analysis and neural network. In the conclusion of the thesis these two approaches and their resulting calculations are compared.