Using least squares support vector machines to the product cost estimation

This research makes the first attempt to apply a novel machine learning method, the least squares support vector machines (LS-SVM), to solving product cost estimation problems in the product life cycle. Four real product cost estimation problems, proposed in previous studies, are used and the estimation performance of LS-SVM model evaluated. These cases include estimations of the costs of carbon steel pipe material, steel pipe bending, pressure vessel manufacturing, and pump purchasing. The performance of numerous cost estimation models, including regression analysis, neural networks, and support vector regression, established in the previous articles, are compared with that of the LS-SVM model. The test results verified that the LS-SVM model can provide more accurate estimation performance and outperforms other methods. The results of this analysis can serve as a useful reference for product cost planning and control in industries. Copyright © 2009 Chung Cheng Institute of Technology.