No description available.

No description available.

Ng Anny. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2002. / Includes bibliographical references (leaves 139-142). / Abstracts in English and Chinese. / Chapter 1 --- Introduction --- p.1 / Chapter 1.1 --- Problem Definition --- p.2 / Chapter 1.2 --- Thesis Organization --- p.3 / Chapter 2 --- Chinese Text Summarization Using Genetic Algorithm --- p.4 / Chapter 2.1 --- Introduction --- p.4 / Chapter 2.2 --- Related Work --- p.6 / Chapter 2.3 --- Genetic Algorithm Approach --- p.10 / Chapter 2.3.1 --- Fitness Function --- p.11 / Chapter 2.3.2 --- Genetic operators --- p.14 / Chapter 2.4 --- Implementation Details --- p.15 / Chapter 2.5 --- Experimental results --- p.19 / Chapter 2.6 --- Limitations and Future Work --- p.24 / Chapter 2.7 --- Conclusion --- p.26 / Chapter 3 --- Event Extraction from Chinese Financial News --- p.27 / Chapter 3.1 --- Introduction --- p.28 / Chapter 3.2 --- Method --- p.29 / Chapter 3.2.1 --- Data Set Preparation --- p.29 / Chapter 3.2.2 --- Positive Word --- p.30 / Chapter 3.2.3 --- Negative Word --- p.31 / Chapter 3.2.4 --- Window --- p.31 / Chapter 3.2.5 --- Event Extraction --- p.32 / Chapter 3.3 --- System Overview --- p.33 / Chapter 3.4 --- Implementation --- p.33 / Chapter 3.4.1 --- Event Type and Positive Word --- p.34 / Chapter 3.4.2 --- Company Name --- p.34 / Chapter 3.4.3 --- Negative Word --- p.36 / Chapter 3.4.4 --- Event Extraction --- p.37 / Chapter 3.5 --- Stock Database --- p.38 / Chapter 3.5.1 --- Stock Movements --- p.39 / Chapter 3.5.2 --- Implementation --- p.39 / Chapter 3.5.3 --- Stock Database Transformation --- p.39 / Chapter 3.6 --- Performance Evaluation --- p.40 / Chapter 3.6.1 --- Performance measures --- p.40 / Chapter 3.6.2 --- Evaluation --- p.41 / Chapter 3.7 --- Conclusion --- p.45 / Chapter 4 --- Mining Frequent Episodes --- p.46 / Chapter 4.1 --- Introduction --- p.46 / Chapter 4.1.1 --- Definitions --- p.48 / Chapter 4.2 --- Related Work --- p.50 / Chapter 4.3 --- Double-Part Event Tree for the database --- p.56 / Chapter 4.3.1 --- Complexity of tree construction --- p.62 / Chapter 4.4 --- Mining Frequent Episodes with the DE-tree --- p.63 / Chapter 4.4.1 --- Conditional Event Trees --- p.66 / Chapter 4.4.2 --- Single Path Conditional Event Tree --- p.67 / Chapter 4.4.3 --- Complexity of Mining Frequent Episodes with DE-Tree --- p.67 / Chapter 4.4.4 --- An Example --- p.68 / Chapter 4.4.5 --- Completeness of finding frequent episodes --- p.71 / Chapter 4.5 --- Implementation of DE-Tree --- p.71 / Chapter 4.6 --- Method 2: Node-List Event Tree --- p.76 / Chapter 4.6.1 --- Tree construction --- p.79 / Chapter 4.6.2 --- Order of Position Bits --- p.83 / Chapter 4.7 --- Implementation of NE-tree construction --- p.84 / Chapter 4.7.1 --- Complexity of NE-Tree Construction --- p.86 / Chapter 4.8 --- Mining Frequent Episodes with NE-tree --- p.87 / Chapter 4.8.1 --- Conditional NE-Tree --- p.87 / Chapter 4.8.2 --- Single Path Conditional NE-Tree --- p.88 / Chapter 4.8.3 --- Complexity of Mining Frequent Episodes with NE-Tree --- p.89 / Chapter 4.8.4 --- An Example --- p.89 / Chapter 4.9 --- Performance evaluation --- p.91 / Chapter 4.9.1 --- Synthetic data --- p.91 / Chapter 4.9.2 --- Real data --- p.99 / Chapter 4.10 --- Conclusion --- p.103 / Chapter 5 --- Mining N-most Interesting Episodes --- p.104 / Chapter 5.1 --- Introduction --- p.105 / Chapter 5.2 --- Method --- p.106 / Chapter 5.2.1 --- Threshold Improvement --- p.108 / Chapter 5.2.2 --- Pseudocode --- p.112 / Chapter 5.3 --- Experimental Results --- p.112 / Chapter 5.3.1 --- Synthetic Data --- p.113 / Chapter 5.3.2 --- Real Data --- p.119 / Chapter 5.4 --- Conclusion --- p.121 / Chapter 6 --- Mining Frequent Episodes with Event Constraints --- p.122 / Chapter 6.1 --- Introduction --- p.122 / Chapter 6.2 --- Method --- p.123 / Chapter 6.3 --- Experimental Results --- p.125 / Chapter 6.3.1 --- Synthetic Data --- p.126 / Chapter 6.3.2 --- Real Data --- p.129 / Chapter 6.4 --- Conclusion --- p.131 / Chapter 7 --- Conclusion --- p.133 / Chapter A --- Test Cases --- p.135 / Chapter A.1 --- Text 1 --- p.135 / Chapter A.2 --- Text 2 --- p.137 / Bibliography --- p.139

M.Ing. (Mechanical Engineering) / The combination of non-linear signal processing and financial market forecasting is a relatively new field of research. This dissertation concerns the forecasting of shares quoted on the Johannesburg Stock Exchange by using Artificial Neural Networks, and does so by comparing neural network results with established statistical results. The share price rise or fall are predicted as well as buy, sell and hold signals and compared to Time Series model and Moving Average Convergence Divergence results. The dissertation will show that artificial neural networks predict the share price rise or fall with less error than statistical models and yielded the highest profit when forecasting buy, sell and hold signals for a particular share.

D.Phil. (Electrical and Electronic Engineering) / Load forecasting is a necessary and an important task for both the electrical consumer and electrical supplier. Whilst many studies emphasize the importance of determining the future demand, few papers address both the forecasting algorithm and computational resources needed to offer a turnkey solution to address the load forecasting problem. The major contribution that, this paper identified is a turnkey load forecasting algorithm. A turnkey forecasting solution is defined by a comprehensive solution that incorporates both the algorithm and processing elements needed to execute the algorithm in the most effective and efficient manner. An electrical consumer, namely the operator of a rapid railway system was faced with a problem of having to forecast the notified network demand and energy consumption. The forecast period was expected to be between a very short term window for maintenance reasons and long term for the requirements warranted by the electrical supplier. The problem was addressed by firstly reviewing the most common forms of load forecasting for which there are two types. These are statistically based methods and methods based upon artificial intelligence. The basic principle of a statistical approach is to approximate or define a curve that best defines the relationship between the load and its parameters. Regression and similar day approach methods use the defined correlation of past values in order to forecast the future behaviour. In other words the future load forecast is forecasted by observing the behaviour of the factors that influenced the load behaviour in the past. The underlying factors that influence the final load may be identified by means of a top down drill down approach. In this way both the load factors and influential variables may be identified. This paper makes use of relevance trees to create a structure of load and influential variables. For a regression forecasting model, the behaviour of the load is modelled according to weather and non-weather variables. The load may be stochastic or deterministic, linear or nonlinear. One of the biggest problems with statistical models is the lack of generality. One model may yield more acceptable results over another model simply because of the sensitivity of the model to one load element that defines the model significantly. Regression type forecast models are an example of this where the elements that define the load are broadly divided into weather and non-weather elements. It is important that the correlation curve reflects the true correlation between the load and its elements. The recursive properties of a statistical based techniques (Kalman filter) allows that the relationship be refined. For methods such as neural networks, the relationship between the load elements that define the future load behaviour is learnt by presenting a series of patterns and then a forecast model is derived. Rigorous mathematical equations are replaced with an artificial neural network where the load curve is learnt. Unlike a statistical based approach (ARMA models), the load does not first need to be defined as a stochastic or deterministic series. In terms of a stochastic approach (non stationery process), the load first would have to be brought to a stationery process. For artificial neural networks, such processes are eliminated and the future forecast is derived faster in terms of a turnkey approach (tested solution). Artificial Neural Networks (ANN) has gained momentum since the eighties. Specifically in the area of forecasting, neural networks have become a common application. In this thesis, data from a railway operator was used to train the neural network and then future data is forecasted. Two embedded processing elements were then evaluated in terms of speed, memory and ability to execute complex mathematical functions (libraries). These were namely a Complex Programmable Logic Device (CPLD) and microcontroller (MCU). The ANN forecasting algorithm was programmed on both a MCU and PLD and compared by means of timing models and hardware platform testing. The most ideal turnkey solution was found to be the ANN algorithm residing on a PLD. The accuracy and speed results surpassed that of a MCU.

Policy makers, water managers, farmers and many other sectors of the society in southern Africa are confronting increasingly complex decisions as a result of the marked day-to-day, intra-seasonal and inter-annual variability of climate. Hence, forecasts of hydro-climatic variables with lead times of days to seasons ahead are becoming increasingly important to them in making more informed risk-based management decisions. With improved representations of atmospheric processes and advances in computer technology, a major improvement has been made by institutions such as the South African Weather Service, the University of Pretoria and the University of Cape Town in forecasting southern Africa’s weather at short lead times and its various climatic statistics for longer time ranges. In spite of these improvements, the operational utility of weather and climate forecasts, especially in agricultural and water management decision making, is still limited. This is so mainly because of a lack of reliability in their accuracy and the fact that they are not suited directly to the requirements of agrohydrological models with respect to their spatial and temporal scales and formats. As a result, the need has arisen to develop a GIS based framework in which the “translation” of weather and climate forecasts into more tangible agrohydrological forecasts such as streamflows, reservoir levels or crop yields is facilitated for enhanced economic, environmental and societal decision making over southern Africa in general, and in selected catchments in particular. This study focuses on the development of such a framework. As a precursor to describing and evaluating this framework, however, one important objective was to review the potential impacts of climate variability on water resources and agriculture, as well as assessing current approaches to managing climate variability and minimising risks from a hydrological perspective. With the aim of understanding the broad range of forecasting systems, the review was extended to the current state of hydro-climatic forecasting techniques and their potential applications in order to reduce vulnerability in the management of water resources and agricultural systems. This was followed by a brief review of some challenges and approaches to maximising benefits from these hydro-climatic forecasts. A GIS based framework has been developed to serve as an aid to process all the computations required to translate near real time rainfall fields estimated by remotely sensed tools, as well as daily rainfall forecasts with a range of lead times provided by Numerical Weather Prediction (NWP) models into daily quantitative values which are suitable for application with hydrological or crop models. Another major component of the framework was the development of two methodologies, viz. the Historical Sequence Method and the Ensemble Re-ordering Based Method for the translation of a triplet of categorical monthly and seasonal rainfall forecasts (i.e. Above, Near and Below Normal) into daily quantitative values, as such a triplet of probabilities cannot be applied in its original published form into hydrological/crop models which operate on a daily time step. The outputs of various near real time observations, of weather and climate models, as well as of downscaling methodologies were evaluated against observations in the Mgeni catchment in KwaZulu-Natal, South Africa, both in terms of rainfall characteristics as well as of streamflows simulated with the daily time step ACRU model. A comparative study of rainfall derived from daily reporting raingauges, ground based radars, satellites and merged fields indicated that the raingauge and merged rainfall fields displayed relatively realistic results and they may be used to simulate the “now state” of a catchment at the beginning of a forecast period. The performance of three NWP models, viz. the C-CAM, UM and NCEP-MRF, were found to vary from one event to another. However, the C-CAM model showed a general tendency of under-estimation whereas the UM and NCEP-MRF models suffered from significant over-estimation of the summer rainfall over the Mgeni catchment. Ensembles of simulated streamflows with the ACRU model using ensembles of rainfalls derived from both the Historical Sequence Method and the Ensemble Re-ordering Based Method showed reasonably good results for most of the selected months and seasons for which they were tested, which indicates that the two methods of transforming categorical seasonal forecasts into ensembles of daily quantitative rainfall values are useful for various agrohydrological applications in South Africa and possibly elsewhere. The use of the Ensemble Re-ordering Based Method was also found to be quite effective in generating the transitional probabilities of rain days and dry days as well as the persistence of dry and wet spells within forecast cycles, all of which are important in the evaluation and forecasting of streamflows and crop yields, as well as droughts and floods. Finally, future areas of research which could facilitate the practical implementation of the framework were identified. / Thesis (Ph.D.)-University of KwaZulu-Natal, Pietermaritzburg, 2007.

by Lam King Wan. / Thesis (M.B.A.)--Chinese University of Hong Kong, 1999. / Includes bibliographical references (leaves 72-73). / ABSTRACT --- p.ii. / TABLE OF CONTENTS --- p.iv. / Section / Chapter I. --- OBJECTIVE --- p.1 / Chapter II. --- INTRODUCTION --- p.3 / The principal investment risk --- p.3 / Effect of risk on investment --- p.4 / Investors' concern for investment risk --- p.6 / Chapter III. --- THE INPUT PARAMETERS --- p.9 / Chapter IV. --- LITERATURE REVIEW --- p.15 / What is an artificial neural network? --- p.15 / What is a neuron? --- p.16 / Biological versus artificial neuron --- p.16 / Operation of a neural network --- p.17 / Neural network paradigm --- p.20 / Feedforward as the most suitable form of neural network --- p.22 / Capability of neural network --- p.23 / The learning process --- p.25 / Testing the network --- p.29 / Neural network computing --- p.29 / Neural network versus conventional computer --- p.30 / Neural network versus a knowledge based system --- p.32 / Strength of neural network --- p.34 / Weaknesses of neural network --- p.35 / Chapter V. --- NEURAL NETWORK AS A TOOL FOR INVESTMENT ANALYSIS --- p.38 / Neural network in financial applications --- p.38 / Trading in the stock market --- p.41 / Why neural network could outperform in the stock market? --- p.43 / Applications of neural network --- p.45 / Chapter VI. --- BUILDING THE NEURAL NETWORK MODEL --- p.47 / Implementation process --- p.48 / Step 1´ؤ Problem specification --- p.49 / Step 2 ´ؤ Data collection --- p.51 / Step 3 ´ؤ Data analysis and transformation --- p.55 / Step 4 ´ؤ Training data set extraction --- p.58 / Step 5 ´ؤ Selection of network architecture --- p.60 / Step 6 ´ؤ Selection of training algorithm --- p.62 / Step 7 ´ؤ Training the network --- p.64 / Step 8 ´ؤ Model deployment --- p.65 / Chapter 7 --- RESULT AND CONCLUSION --- p.67 / Result --- p.67 / Conclusion --- p.69 / BIBLIOGRAPHY --- p.72

This thesis consists of three parts: a genetic fuzzy trading model for stock trading, incremental intraday information for financial time series forecasting, and intraday effects in conditional variance estimation. Part A investigates a genetic fuzzy trading model for stock trading. This part contributes to use a fuzzy trading model to eliminate undesirable discontinuities, incorporate vague trading rules into the trading model and use genetic algorithm to select an optimal trading ruleset. Technical indicators are used to monitor the stock price movement and assist practitioners to set up trading rules to make buy-sell decision. Although some trading rules have a clear buy-sell signal, the signals are always detected with 'hard' logical. These trigger the undesirable discontinuities due to the jumps of the Boolean variables that may occur for small changes of the technical indicator. Some trading rules are vague and conflicting. They are difficult to incorporate into the trading system while they possess significant market information. Various performance comparisons such as total return, maximum drawdown and profit-loss ratios among different trading strategies were examined. Genetic fuzzy trading model always gave moderate performance. Part B studies and contributes to the literature that focuses on the forecasting of daily financial time series using intraday information. Conventional daily forecast always focuses on the use of lagged daily information up to the last market close while neglecting intraday information from the last market close to current time. Such intraday information are referred to incremental intraday information. They can improve prediction accuracy not only at a particular instant but also with the intraday time when an appropriate predictor is derived from such information. These are demonstrated in two forecasting examples, predictions of daily high and range-based volatility, using linear regression and Neural Network forecasters. Neural Network forecaster possesses a stronger causal effect of incremental intraday information on the predictand. Predictability can be estimated by a correlation without conducting any forecast. Part C explores intraday effects in conditional variance estimation. This contributes to the literature that focuses on conditional variance estimation with the intraday effects. Conventional GARCH volatility is formulated with an additive-error mean equation for daily return and an autoregressive moving-average specification for its conditional variance. However, the intra-daily information doesn't include in the conditional variance while it should has implication on the daily variance. Using Engle's multiplicative-error model formulation, range-based volatility is proposed as an intraday proxy for several GARCH frameworks. The impact of significant changes in intraday data is reflected in the MEM-GARCH variance. For some frameworks, it is possible to use lagged values of range-based volatility to delay the intraday effects in the conditional variance equation. / Ng, Hoi Shing Raymond. / Adviser: Kai-Pui Lam. / Source: Dissertation Abstracts International, Volume: 72-01, Section: B, page: . / Thesis (Ph.D.)--Chinese University of Hong Kong, 2010. / Includes bibliographical references (leaves 107-114). / Electronic reproduction. Hong Kong : Chinese University of Hong Kong, [2012] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Electronic reproduction. Ann Arbor, MI : ProQuest Information and Learning Company, [200-] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Abstract also in Chinese.

by Ng Wai-Leung Anthony. / Thesis (M.Phil.)--Chinese University of Hong Kong, 1999. / Includes bibliographical references (leaves 83-87). / Abstracts in English and Chinese. / Chapter I. --- INTRODUCTION --- p.1 / Chapter II. --- LITERATURE REVIEW --- p.5 / Chapter 2.1 --- The Importance of Stock Index and Index Futures --- p.6 / Chapter 2.2 --- Importance of Index Forecasting --- p.6 / Chapter 2.3 --- Reasons for the Lead-Lag Relationship between Stock and Futures Markets --- p.9 / Chapter 2.4 --- Importance of the lead-lag relationship --- p.10 / Chapter 2.5 --- Some Empirical Findings of the Lead-Lag Relationship --- p.10 / Chapter 2.6 --- New Approach to Financial Forecasting - Artificial Neural Network --- p.12 / Chapter 2.7 --- Artificial Neural Network Architecture --- p.14 / Chapter 2.8 --- Evidence on the Employment of ANN in Financial Analysis --- p.20 / Chapter 2.9 --- Hong Kong Securities and Futures Markets --- p.25 / Chapter III. --- GENERAL GUIDELINE IN DESIGNING AN ARTIFICIAL NEURAL NETWORK FORECASTING MODEL --- p.28 / Chapter 3.1 --- Procedure for using Artificial Neural Network --- p.29 / Chapter IV. --- METHODOLOGY --- p.37 / Chapter 4.1 --- ADF Test for Unit Root --- p.38 / Chapter 4.2 --- "Error Correction Model, Error Correction Model with Short- term Dynamics, and ANN Models for Comparisons" --- p.38 / Chapter 4.3 --- Comparison Criteria of Different Models --- p.39 / Chapter 4.4 --- Data Analysis --- p.39 / Chapter 4.5 --- Data Manipulations --- p.41 / Chapter V. --- RESULTS --- p.42 / Chapter 5.1 --- The Resulting Models --- p.42 / Chapter 5.2 --- The Prediction Power among the Models --- p.45 / Chapter 5.3 --- ANN Model of Input Variable Selection Using Contribution Factor --- p.46 / Chapter VI. --- CAUSALITY ANALYSIS --- p.54 / Chapter 6.1 --- Granger Casuality Analysis --- p.55 / Chapter 6.2 --- Results Interpretation --- p.56 / Chapter VII --- CONSISTENCE VALIDATION --- p.61 / Chapter VIII --- ARTIFICIAL NEURAL NETWORK TRADING SYSTEM --- p.67 / Chapter 7.1 --- Trading System Architecture --- p.68 / Chapter 7.2 --- Simulation Runs using the Trading System --- p.77 / Chapter XI. --- CONCLUSIONS AND FUTURE WORKS --- p.79