Graham Kendall

Individual Projects (3rd Year Undergraduates)

(Academic Year 2003/04)

Introduction

This page details the third year individual projects that are/have been supervised by Graham Kendall who is a lecturer in the Department of Computer Science at The University of Nottingham. This page contains details of projects supervised in the academic year 2003/04.

The main project page can be found here.

Downloadable Dissertations

Some of the projects are available for download, so that you can view them. This facility is for current third year students, being supervised by me, who can use previous dissertations for their own research and also to look at previous dissertations to see how to lay them out etc. (it also saves the inevitable requests I receive asking to see a "good" project from last year, with the risk that I never see it again!).

As such, these files are password protected and, to get the username/password you need to EMAIL me. People outside of the university will not be given access.

In addition, I cannot give access to those students outside of my tutor group. You should approach your own tutor to get copies of previous dissertations they have supervised.

Note : I cannot make the files available in formats other than those already supplied.


3rd Year Projects Supervised in 2003/04

  1. Predicting Equity Movements using a Neural Network by Andrew Diprose (PDF : 1,592KB)
    Abstract : With more advanced technology, investors have quicker and more reliable access to the equity market than ever before. This has the effect of making it more volatile and seemingly random. I believe that there is method to this madness, and that with the right methods, movements in the equity market can be predicted. To do this, I implemented a multi-layer, feed-forward, back-propagation neural network that used a log-sigmoid transfer function. After programming my own Neural Network that is capable of solving the XOR, I decided to implement a neural network within an already tried, tested and optimised neural network package, MATLAB; feeding it PYTHON pre-processed real historic equity data extracted using DATASTREAM. After extensive testing and streamlining of the neural network, I had a very efficient system capable of predicting equity market movements correctly 70% of the time. Simulating this system using random historic data, I was able to calculate it was capable of making on average £422.28 profit a day, which equates to £105,147.72 a year, using initial investments of £1000. The relationship between initial investments and profit is linear, allowing for even greater profit given a larger initial capital base.


  2. Fixture Scheduling for the English Premiership by Amjid Raja (PDF: 670KB)
    Abstract : Real life sports scheduling problems are very difficult to solve due to a multitude of often conflicting requirements and preferences. In this dissertation I will discuss a scheduling process for the English F.A. Premiership which attempts to minimise manual intervention whilst producing a fixture schedule that satisfies all requirements and also reduces the number of breaks in the home and away patterns for clubs. The Premiership relies heavily on a fair schedule and so most requirements are treated very seriously. I will investigate the work of other researchers in the realm of fixture scheduling for sports leagues looking closely at approaches covering round robin
    tournaments.
    My aim is to produce a schedule for the Premiership season 2003/2004 which satisfies all the constraints and reduces breaks whilst minimising the manual effort.

  3. Sports Fixture Scheduling by Nitesh Modha (PDF: 229KB)
    Abstract : Sports leagues have to deal with many problems when scheduling tournaments. These problems contain many conflicting constraints, which are usually brought up by the teams, television and media, fans, police and their governing bodies. This paper deals with the problems of scheduling fixtures for the four top English Leagues: The Premier League, Division 1, Division 2 and Division 3. I have used a heuristic approach with a simulated annealing technique to successfully create my schedule, using a multiple step process.

  4. Investigation of the Construction and Optimisation of the Football League by Robert Lee (PDF: 845KB) (Word: 622KB)
    Abstract : This dissertation investigates a way of providing a suitable schedule for the English football league. In general scheduling deals with the allocation of resources over time to perform a collection of tasks. In football scheduling, fixtures are allocated to time slots under a number of constraints.
    Fixture scheduling in the English league plays an important part into the potential revenue of football and influences the general interest of the football watchers. It is a very complex task due to the many conflicting requirements and preferences from various organisations and the fans.
    There has been a fair amount of work on sports scheduling including successful implementations in basketball. However there has been less work in the field of football scheduling and the fact that football tends to be more highly constrained makes it a more challenging problem. There are common features that exist amongst almost all sports such as home and away patterns which suggest that similar approaches from the other sports can be used in football.
    To solve this problem a constructive heuristic was developed to construct a basic schedule. The heuristic was implemented in a variety of ways to vary the feasibility of the schedule. The assessment of the quality of a schedule was done using an evaluation function. This penalised the schedule where constraints were broken with different weights being given for stronger and weaker constraints. To improve upon the initial schedule an optimiser was created. This was used to test how great an improvement could be made from different quality initial schedules. The optimised schedule was then compared to a real life schedule and evaluations were made from this comparison to determine the quality of the system.
  5. An Agent Architecture For A Simulated Android Judo Combatant by Dominic Price (PDF: 1,212KB)
    Abstract : There has been a great deal of interest in the press in recent times about the advances made in designing and building humanoid robots, in particular the developments made by Sony with their QRIO robot. With this in mind an international contest was announced in July 2003 where the aim was to design a piece of software to control a virtual robot running in a 3D simulator, and to have that robot controller compete against others in a game of Robot Judo. I entered the contest in November 2003 with only a little knowledge of the field but with the belief that I could design and implement a controller that would be able to compete at this level. I first designed a controller with a reactive agent architecture that would be able to move, fight and track the opponent visually.
    In implementing this design I encountered a few problems but the controller was still effective and the highest ranking that I achieved during the contest was rank 7, and the finishing position was rank 21 among 52 competitors.

  6. The Application of Neural Network Techniques to Horse Race Prediction by Adam Stringer (PDF: 829KB) (Word: 3,759KB)
    Abstract : Research on the use of artificial intelligence systems for solving complex problems has led to the development of a number of techniques and applications. Such systems attempt to model the human brain structure, making them suited to non-linear, data intensive problems that were previously difficult to compute. The UK horse racing market is an extremely profitable business for those involved in the industry, however gambling on the horses is a much more risky prospect. This project introduces the concept of the horse racing market and the efficient market hypothesis. By evaluating existing artificial intelligence techniques, a suitable system is designed and implemented for the purpose of researching horse race prediction. An analysis of the results provides evidence about the efficiency of the market and the scope for profitability.

  7. Poker Companion by Michael Lane (PDF: 5,384KB) (Word: 15,345KB)
    Abstract : Automated poker players have been developed in the past but had not been very successful due to the amount of strategy needed to play poker to a high level. Each human player will play with an entirely different strategy to someone else depending on their circumstances. This is also true for different types of poker table; on a free play table players will bet as high as possible on any remotely good hand because they didnt pay for their chips in the first place and they dont care if all playing chips are lost, where as players are likely to be more prudent when their own money is at risk.
    My dissertation describes a poker companion which I have designed and programmed. It is a stand alone program which runs in the background while someone is playing poker.
    If the companion is running while the user is paying an online game of Texas Holdem poker at a compatible table, like those at www.ladbrokes.com, it will give them a percentage chance of winning the hand at the two most important stages of betting. It provides a probability of winning the hand depending on the hole cards the player has drawn, it also gives valuable information to the player telling them what the probability of them winning the hand is depending on past experiences.
    The method of creating these probabilities is loosely base around the theory of case based reasoning and simple self evolution. This being the case my program will improve as it is used and give the users reliable feedback to improve their chance of winning at an online poker table.

  8. Scheduling the English Football League's Boxing Day Round of Fixtures by Andrew Tucker (Word: 849KB)
    Abstract : This study aimed to overcome the specific constraints for scheduling the English Football Leagues Boxing Day round of fixtures. The main objective was to produce solutions of a higher quality than those published by the League in recent seasons. The constraints presented by the problem are similar to those that have been considered in recent research in the area of sports scheduling. Of principal concern is the requirement that the travelling distances involved for each fixture are minimised. During the course of this study, local search techniques were investigated in order to implement a constructive heuristic together with a randomised local search. A successful system was produced that generated significantly improved results in comparison to the published fixtures.

  9. Designing an application that assists online poker players to play their hole cards to optimum value by Niall Charlton (Word: 4,212KB)
    Abstract : This dissertation describes an application that will provide poker players with vital information to assist them with their decision-making. Through an interface the user will be able to select what cards they have, how many players are in the round and the board cards. From this my program will work out vital pre-flop information that will help users decide upon their action. The application will improve online poker player's play by directing them to play the correct starting hands in the correct position.

  10. Implementing a Scrabble Move Generator Using Partial-Match Algorithm by Sheak Yin Sim (PDF: 844KB)
    Abstract : This dissertation investigates, and examines, the implementation of partial-match search algorithm and the feasibility of this algorithm as an alternative approach to implement move generator (i.e. finding valid words to play) in a Scrabble game. Various past works that have been done on this algorithm is reviewed and the most suitable algorithm an algorithm using Ternary Search Tree is selected to solve the problem. Firstly, proof of concept is conducted in order to show that the algorithm does work, given a small collection of words. Following the proof of concept stage, an experiment is performed on a complete dictionary. This is to examine if the implementation is able to support a large dictionary. In the final stage, the algorithm is incorporated into the move generator for a Scrabble game, which aims to generate the highest scoring word to play for a given board layout and tiles rack.

  11. Sports Scheduling in Round Robin Competitions by Peter Bloor (PDF: 86KB)
    Abstract : Creating sports schedules by hand is an arduous task. It becomes extremely difficult when the number of teams involved is large, especially when there are numerous requirements and constraints involved. This project creates an automated scheduling system for numerous individual round robin leagues, aiming to place them all in one final schedule.

  12. Ultimate Frisbee Scheduling by James Cooper (Word : 7,981KB)
    Abstract : Ultimate Frisbee is a fast-paced team sport that can be played indoors or outdoors. Ultimate Frisbee differs from many traditional sports scheduling problems. Tournaments are typically run over a weekend, with the schedule needing to accommodate up to forty teams. Information often changes at the last-minute and this can lead to the schedule having to be recalculated with little notice. This system is a website that offers Ultimate-style schedules, based upon recommendations from real life Tournament Directors and the standards set by the U.K. governing body. PHP and MYSQL will be used to implement this project, since they allow the fulfilment of many of the proposed recommendations. My system will produce the schedule that, with other information pertaining to the tournament, forms a "Captains' Pack": a complete guidebook for the whole tournament. The system focuses on usability and accessibility by those without computer knowledge, and produces the schedule extremely quickly. Several recommendations are laid out for continuing the project into a more complete resource.

 


EMAIL : gxk@cs.nott.ac.uk          Home page

Last Updated 15th June 2001