Inference of genetic networks using random forests:Performance improvement using a new variable importance measure

Among the various methods so far proposed for genetic network inference, this study focuses on random-forest-based methods. Confidence values are assigned to all of candidate regulations when taking approach. To our knowledge, make assignments using standard variable importance measure defined in tree-based machine learning techniques. Therefore, sum confidence a certain gene from other genes, that computed single random forest, is always restricted value almost 1. We think feature inconvenient inference requires compare multiple forests. In we therefore propose an alternative measure, what call ``the random-input measure,'' and design new method uses place existing method. show, through numerical experiments, use improves performance by as much 45.5% with respect area under recall-precision curve (AURPC).