Remanufacturability Evaluation Method for Used Vehicles Based on Stacking Ensemble Learning Framework

Abstract

This paper proposes a novel method for evaluating the remanufacturability of used vehicles based on Stacking-Based Ensemble Learning Algorithm (SBELA). A method that combines a supply chain evolutionary game model is proposed to construct a Hybrid Dataset (HD), which aims to deal with Remanufacturing Data Gap (RDG). A modified SBELA is used for evaluating the remanufacturability of used vehicles. The results of this investigation show that HD can be used to evaluate remanufacturability in the initial stage of remanufacturing. The modified SBELA significantly improves the evaluation of remanufacturability performance, compared to the Ridge Regression, Lasso Regression, Linear Regression, Stochastic Gradient Descent (SGD), Kneighbors, AdaBoost, and Gradient Boosting Regression (GBR), the Mean Square Error (MSE) has decreased by 51.88%, 53.75%, 52.05%, 58.15%, 8.33%, 57.92%, and 22.22%, respectively. The investigation results demonstrate HD’s effectiveness in evaluating remanufacturability during the initial remanufacturing stage. This can provide a reference for newly established remanufacturing enterprises in the RDG dilemma.

Keywords

• Stacking ensemble learning
• evaluation of remanufacturability
• hybrid dataset
• remanufacturing data gap