Year of Graduation
Program for feature selection by precedents using GLMNET algorithm
School of Software Engineering
In this paper we consider an algorithm for feature selection for the machine learning problems. The current problem is a part of the amount of methods from dimensionality reduction field, which helps the developer to boost the source data quality and the machine learning algorithm quality itself. Feature selection problem has a high topicality because of popularity of the machine learning algorithms in those areas of industry and applied science where information technologies are used. According to the specifications we develop the feature selection method using GLMNET algorithm. During the development we perform the analisys of the subject area, the existing feature selection methods and the adjacent concepts. Also we implement the developed method and perform its testing on the applied problem. Key words: glmnet, elastic-net, regularization, logistic regression, machine learning, feature selection.