A Constructive Induction of Feature using Random Forest Approach
Wiesław Paja, Maksymilian Knap · Procedia Computer Science · 2020
Machine learning, including the selection of important features, has a very wide and important application in medicine. Feature selection is an important stage in the analysis of medical data and aims to simplify the problem, reduce dimensionality but also to identify important parameters, symptoms that are important in diagnosing diseases. The presented approach is a kind of concept for automation of the process of determining the relationship between the attributes describing the problem. It can also be a way to define features that are interdependent. In this way it is possible to identify redundant features existing in the dataset. The method is based on the construction of new features based on the analysis of the relevance of the original features. The initial results obtained are an interesting introduction to further extensive research on constructive induction of feature.