An obsolescence forecasting method based on improved radial basis function neural network

With the rapid development of technologies, product lifecycle becomes shorter, which brings great challenges to obsolescence management. An efficient obsolescence forecasting method is in need. This research proposes a two-stage obsolescence forecasting model. The first stage identifies the key product features for obsolescence with ELECTRE I method. The second stage calculates the obsolescence probability based on radial basis function neural network. Three improvements are made for better predication accuracy, (1) information gain and information gain ratio are integrated to calculate the input weights; (2) an improved Particle Swarm Optimization algorithm is applied to calculate the clustering centroid; and (3) an improved gradient descent method determines the weights from hidden layer to output layer. The performance of the proposed method is compared with the existing model by using mobile dataset which contains 7000 samples. The experimental result shows that the accuracy of the predication has been improved from 92.2% to 95.23%. CO 2022 THE AUTHORS. Published by Elsevier BV on behalf of Faculty of Engineering, Ain Shams University This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-ncnd/4.0/).