Modeling and Forecasting of High-Technology Manufacturing Labor Productivity Based on Grey Support Vector Machines with Genetic Algorithms

Abstract

In recent years, computing high-technology manufacturing (HTM) labor productivity (LP) level and growth rate has gained a renewed interest in both growth economists and trade economists. Measuring LP performance has become an area of concern for companies and policy makers. HTM LP is complex to conduct due to its nonlinearity of influenced factors. Support vector machines (SVM) have been successfully employed to solve nonlinear regression and time series problems. Grey system theory successfully utilizes accumulated generating data instead of original data to build forecasting model, which makes raw data stochastic weak, or reduces noise influence in a certain extent. However, the application combining grey system theory and SVM for LP forecasting is rare. In this study, a grey support vector machines with genetic algorithms (GSVMG) is proposed to forecast HTM LP. In addition, GM (1, N) model of grey system is used to add a grey layer before neural input layer and white layer after SVM layer. Genetic algorithms (GAs) are used to determine free parameters of support vector machines. Evaluation method has been used for comparing the performance of forecasting techniques. The experiments show that the GSVMG model is outperformed GM (1, N) model and SVM with genetic algorithms (SVMG) model, and HTM LP forecasting based on GSVMG is of validity and feasibility.

Grey SVM with Simulated Annealing Algorithms in Patent Application Filings Forecasting

“…The superior performance of the GSVMS model has several causes. First, the GSVMS model exploits sufficiently the characteristic of the preprocessed data handled by the grey operation with stochastic reduced and regularity raised and the nonlinear map feature of SVM, which makes the convergent process of the SVM fast [10] [11]. Second, improper determining of these three parameters will cause either over-fitting or under-fitting of a SVM model.…”