<![CDATA[Demand forecasting for port critical spare parts (CSP) is notoriously difficult as it is expensive, lumpy and intermittent with high variability. In this paper, some influential factors which have an effect on CSP consumption were proposed according to port CSP characteristics and historical data. And analytic hierarchy process (AHP) is used to sieve out the more influential factors. Combined with the influential factors, a least squares support vector machines (LS-SVM) model optimized by particle swarm optimization (PSO) was developed to forecast the demand. And the effectiveness of the model is demonstrated through a real case study, which shows that the proposed model can forecast the demand of port CSP more accurately, and effectively reduce inventory backlog. DOI : http://dx.doi.org/10.11591/telkomnika.v12i5.3238]]>
Copyrights © 2014
