國立臺灣大學工業工程學研究所Chang, Shi-ChungShi-ChungChangHsu, Chia-HauChia-HauHsuCheng, Yee-ChiuYee-ChiuChengChen, ArgonArgonChenGuo, Ruey-ShanRuey-ShanGuo2006-09-272018-06-292006-09-272018-06-292001http://ntur.lib.ntu.edu.tw//handle/246246/200609271216521641062001 deliverable report:. intelligent multidimensional demand aggregation/disaggregation strategies. task 879.001: intelligent demand aggregation & forecast solutions. project 879: demand data mining & planning in semiconductorThe demand uncertainty propagated and magnified over the semiconductor demand-supply network is the crucial cause of poor manufacturing/logistic plans. To manage the demandvariability, appropriate demand aggregation/forecasting approaches are known to be effective. In the first year of this research task, a multivariate time series model is used as a study vehicle to investigate the effect of aggregating interrelated demands. Heuristic demand grouping algorithms, i.e., a variety of Greedy algorithms, are also developed to minimize the safety stock costs under demand uncertainty. The research results provide practitioners practical guidelines and methodologies to select proper aggregation/forecasting approaches and to group demands for minimum safety stock costs.application/pdf154927 bytesapplication/pdfzh-TWconference paperhttp://ntur.lib.ntu.edu.tw/bitstream/246246/20060927121652164106/1/2001 deliverable report.pdf