GMADM-based attributes selection method in developing prediction model

Abstract

Attribute Selection is an important issue for developing a prediction model, however, how to determine an effective attribute selection algorithm is an important but difficult issue. Attribute selection can effectively delete the irrelevant and redundant attributes to increase the prediction accuracy, and evaluating attribute selection methods usually need to consider several criteria such as accuracy, type I error, and type II error. In this paper, the selected attribute process is modeled as a group multiple attributes decision making (GMADM) problem. In evaluating different GMACD methods, the most results usually are consistently, But there are some situations where the evaluated outcomes have different results. The GMADM method is useful tool for evaluating attribute selection algorithms, and the TOPSIS is capable of identifying a compromised solution when different GMADM method result in conflicting rankings. Therefore, this paper proposes an objective (persuasive) GMADM-based attributes selection method to solve this disagreement and help decision makers pick the most suitable method. After verification, the proposed model is more persuasive to evaluate the attributes selection methods for developing prediction model.