Applications of Data Mining Techniques in Demand Forecasting and Capacity Planning for DRAM Fabrication
碩士 === 國立臺北科技大學 === 工業工程與管理研究所 === 92 === Economic cycle and market factors always affect DRAM price. For DRAM fabrications, much effort has been made to reduce cost. DRAM fabrications allocate capacity at the right time and produce product that meets market demand, which will both increase profit a...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | zh-TW |
| Published: |
2004
|
| Online Access: | http://ndltd.ncl.edu.tw/handle/14469459979858436436 |
| Summary: | 碩士 === 國立臺北科技大學 === 工業工程與管理研究所 === 92 === Economic cycle and market factors always affect DRAM price. For DRAM fabrications, much effort has been made to reduce cost. DRAM fabrications allocate capacity at the right time and produce product that meets market demand, which will both increase profit and reduce risk. The product mix is based on customers’ order and sales forecast.
In this project, we use data mining techniques to find the relationships between average selling prices and sales demands of DRAM and then plan capacity for front-end and back-end. It can help DRAM manufacturers make the proper production decision. The purposes of this project is to：
1.Correctly forecast the customer demands, and make the decision of wafer releasing.
2.Decide outsourcing strategy of the back-end assembly and final test according to demand forecasting.
3.Deliver customers’ demand on time.
4.Allocate the contractors and spot market based on sales demand and Fab capacity.
5.Monitor DRAM supply chain.
6.Provide top managers information to make decisions.
This research uses Pearson correlation analysis to find the relation of average selling price and sales demand. A sales demand forecasting system based on artificial neural network technology is developed to accurately forecast the future sales demand. It will help DRAM manufacturers make profitable capacity plan and appropriate backend capacity allocation.
|
|---|