Distribution Properties of the Selected Genes 10 12 Median expression level

The expression level matrix resulting from the ANOVA filter is already of manageable size, although we still have many more features than observations. In effect, in our modeling attempts described in Section 5.4, we will apply the selected models to this matrix. Nevertheless, one can question whether better results can be obtained with a dataset with a more “standard” dimensionality. In this context, we can try to further reduce the number of features and then compare the results obtained with the different datasets.