Cuder Gerald, Baumgartner Christian
2018
Cancer is one of the most uprising diseases in our modern society and is defined by an uncontrolled growth of tissue. This growth is caused by mutation on the cellular level. In this thesis, a data-mining workflow was developed to find these responsible genes among thousands of irrelevant ones in three microarray datasets of different cancer types by applying machine learning methods such as classification and gene selection. In this work, four state-of-the-art selection algorithms are compared with a more sophisticated method, termed Stacked-Feature Ranking (SFR), further increasing the discriminatory ability in gene selection.