碩士 / 國防大學中正理工學院 / 兵器系統工程研究所 / 92 / Rapid and correct supplying activities are very important for the repairing system of R.O.C NAVY. For example, lacking spare parts will delay the schedule of maintenance, and too many spare parts which are more than real requirements will waste a lot of money. For these two reasons, it is important to get correct numbers of spare parts before the maintenance started.
In the research a Data Mining technique is used to forecast the requirements of spare parts. We first collect the spare parts used records, and then we establish the Data Mining model by using Data Mining tool named SQL Sever 2000 Analysis Services and the mining algorithms named Decision Tree and Clustering. After being trained by these collected data sets, we get two kinds of models and some hiding rules that can be used to calculate the forecasting amounts of spare parts. Finally, we compare the forecasting numbers of these two models with the real records, and decide which model is better to predict appropriate amount of spare parts.
It is shown in this research that based on the comparison of the results of two models with the real records, we conclude that the result of Decision Tree model is better than the result of Clustering model.
| Identifer | oai:union.ndltd.org:TW/092CCIT0157005 |
| Date | January 2004 |
| Creators | 林聖義 |
| Contributors | 孟興漢 |
| Source Sets | National Digital Library of Theses and Dissertations in Taiwan |
| Language | zh-TW |
| Detected Language | English |
| Type | 學位論文 ; thesis |
| Format | 76 |
Page generated in 0.0142 seconds