Process mining is an emerging research discipline that provides techniques that can be used to discover, monitor and improve real processes using event data. In this thesis we present several approaches and extensions that improve the effectiveness of the Evolutionary Tree Miner, a genetic process mining algorithm. These approaches and extensions enable the Evolutionary Tree Miner to make smart changes to process models, in order to obtain models of a higher quality in less time than the original implementation, while taking into account the four process model quality dimensions of replay fitness, simplicity, precision and generalisation. The approaches and extensions are based on concepts and ideas from process model repair, which have been applied in the context of the Evolutionary Tree Miner. We show, through experiments on both artificial and randomly generated event logs, that our approach is superior to the original implementation of the Evolutionary Tree Miner in its ability to quickly produce high quality model.