Please use this identifier to cite or link to this item:
Title: Design and development of a knowledge discovery system in inventory management
Authors: Mitrea, Catalin
Keywords: DRNTU::Engineering::Manufacturing::Inventory control
Issue Date: 2010
Source: Mitrea, C. (2010). Design and development of a knowledge discovery system in inventory management. Master’s thesis, Nanyang Technological University, Singapore.
Abstract: The ability to learn inductively from examples is an important feature of intelligent systems. These learning methods and algorithms, which are able to generate a model of a system using the observations (data) available, are applied in a specific domain usually with the following goals: to create a system that can carry out a task and to obtain a better understanding of the available data. In this thesis, function approximation or pattern recognition is used to predict inventory level so as to enhance the performance of supply chain. Inventory management plays a crucial role in logistics and its performance is shaped dramatically by uncertainty in both demand and supply. Although demand is difficult to predict due to its stochastic behavior, it is necessary to have accurate forecasting so as to fulfill the customer needs and maintain the corporate competitive edge. Achieving accurate demand forecasting and representing knowledge in rules attract the attention of both academic and practitioners. Knowledge is regarded as a valuable asset for enterprises and knowledge can be manipulated through Artificial Intelligence (AI) techniques. Among those AI techniques, Artificial Neural Networks (ANN) has been proven to be very useful in approximating functions and recognizing patterns by adjusting their weights and biases during training process, but it cannot explain how they arrived at a conclusion in solving a problem. This is the reason why ANN is considered in literature as black boxes due to the lack of clear explanation. To overcome these shortages, TREes PArroting Networks(TREPAN) algorithm is used to extract knowledge from the trained networks in the form of decision trees. The TREPAN algorithm utilizes the IF M-of-N THEN conditions rule instead of the common IF-THEN-ELSE rule to construct the decision tree which can be used to understand previously unknown relationship between input variables for forecasting so as to improve inventory management. The experimental results show that forecasting accuracy using ANN are superior to traditional methods like moving average and autoregressive integrated moving average (ARIMA). Also, the knowledge extracted from trained ANNs using TREPAN algorithm is represented in a comprehensible way and can be used to facilitate human decision-making. The significance of this study is to design and develop a knowledge discovery system, thereby enhancing the inventory management.
DOI: 10.32657/10356/38766
Fulltext Permission: open
Fulltext Availability: With Fulltext
Appears in Collections:MAE Theses

Files in This Item:
File Description SizeFormat 
MitreaCatalin2010.pdfReport1.83 MBAdobe PDFThumbnail

Page view(s) 50

Updated on Aug 5, 2021

Download(s) 20

Updated on Aug 5, 2021

Google ScholarTM




Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.