Created by W.Langdon from gp-bibliography.bib Revision:1.5327
Designing an effective decision-support system for SCM has become crucial in recent years. With the advent of e-Commerce and in a global economy, SCM systems have to be able to deal with the uncertainty and volatility of modern markets. In such systems, the ability to learn and adapt to new conditions in the environment is of paramount importance. This thesis proposes a reusable demand-driven SCM solution. It introduces a number of algorithms for solving different tasks across the supply chain, including: constraint satisfaction, prediction, planning, and online adjustments.
The main contribution of the thesis lies in exploring the problem of dynamic pricing and predicting on line auctions in the context of SCM. A number of algorithms for modelling competitors' behaviour and predicting customer order prices are proposed and compared in the thesis. Their influence on the overall system performance is also explored. The algorithms are based on the Neural Networks and Genetic Programming learning techniques.
The Trading Agent Competition SCM game is used as a testbed for analysing the proposed methods. Although only the results from testing the algorithms in this simulated environment are provided, the methods are not tied to the game rules and can successfully be used for predicting other financial instruments and in other competitive trading environments.",
supervisor: Dr. Maria Fasli",
Genetic Programming entries for Yevgeniya Kovalchuk