Optimization of Obsolescence Forecasting Using New Hybrid Approach Based on the RF Method and the Meta-heuristic Genetic Algorithm

Abstract

Obsolescence is highly complex problems due to the influence of many factors such as technological advancement. However, prediction of obsolescence appears to be one of the most efficient solutions. This paper proposes a novel approach known as GA-RF for obsolescence forecasting. Genetic algorithm (GA) searches for optimal parameters and feature selection to construct a random forest (RF) in order to improve the classification of RF. To examine the feasibility of this approach, this paper presents a comparison between GA-RF, RF, Stepwise logistic regression, and stochastic gradient boosting. Experimental results show that GA-RF outperformed the other methods with 93.3% of accuracy, 90.4% of sensitivity and 95.4% of specificity.