A high performing meta-heuristic for training support vector regression in performance forecasting of supply chain. Issue 8 (November 2016)

• Record Type: Journal Article
• Authors: Vahdani, Behnam; Razavi, Farzad; Mousavi, S.
• Is Part Of: Neural computing & applications. Volume 27:Issue 8(2016)
• Page Start: 2441
• Page End: 2451
• Publication Date: 2016-11
• Subjects: Imperialist competitive algorithm (ICA) -- Support vector regression -- Supply chain -- Cosmetics production; Neural networks (Computer science) -- Periodicals; Neural circuitry -- Periodicals; Artificial intelligence -- Periodicals; Neural Networks (Computer) -- Periodicals; Réseaux neuronaux (Informatique) -- Périodiques; Réseaux nerveux -- Périodiques; Intelligence artificielle -- Périodiques; 006.32
• DOI: 10.1007/s00521-015-2015-8
• Languages: English
• ISSNs: 0941-0643

Abstract

Abstract The prevalence of the use of third-party logistics (3PL) providers is noticeable. The complexity of the relationships pertinent to 3PL is greater than that of any traditional logistics supplier relationships. Moreover, they can be considered as truly strategic alliances. The use of the mentioned relationships to increase the flexibility of the organization to address the rapid changes occurring in market conditions has become popular while these relationships concentrate on the core competencies as well as the development of long-term growth strategies. A good number of studies have examined the selection of service providers. With respect to the selection of the service providers, the most recent studies approved the better performance of neural networks in comparison with the conventional methods to provide a solution for the real-world engineering problems, one of the sociopolitically inspired optimization strategies named imperialist competitive algorithm (ICA) is used. In order to select the 3PL, integration of the support vector regression (SVR) and self-adaptive ICA (SAICA) has offered a novel model, in which SAICA is utilized to adjust the parameters of the SVR. The suggested model is applied for cosmetics production. Moreover, the comparison of the suggested model and back-propagation neural networks, pure SVR, and ICA–SVR is presented. Higher estimation accuracy is achieved as the results of the proposed model reveal, which leads to the effective prediction.