Susan Bedingfield

Extraction of high quality k-words for alignment-free sequence comparison

The weighted Euclidean distance (D(2)) is one of the earliest dissimilarity measures used for alignment free comparison of biological sequences. This distance measure and its variants have been used in numerous applications due to its fast computation, and many variants of it have been subsequently introduced. The D(2) distance measure is based on the count of k-words in the two sequences that are compared. Traditionally, all k-words are compared when computing the distance. In this paper we show that similar accuracy in sequence comparison can be achieved by using a selected subset of k-words. We introduce a term variance based quality measure for identifying the important k-words. We demonstrate the application of the proposed technique in phylogeny reconstruction and show that up to 99% of the k-words can be filtered out for certain datasets, resulting in faster sequence comparison. The paper also presents an exploratory analysis based evaluation of optimal k-word values and discusses the impact of using subsets of k-words in such optimal instances.

A decision tree approach for predicting smokers’ quit intentions

This paper presents a decision tree approach for predicting smokerspsila quit intentions using the data from the International Tobacco Control Four Country Survey. Three rule-based classification models are generated from three data sets using attributes in relation to demographics, warning labels, and smokerspsila beliefs. Both demographic attributes and warning label attributes are important in predicting smokerspsila quit intentions. The modelpsilas ability to predict smokerspsila quit intentions is enhanced, if the attributes regarding smokerspsila internal motivation and beliefs about quitting are included.

Evolutionary rule generation classification and its application to multi-class data

This paper considers an evolutionary algorithm based on an information system for generating classification rules. Custom genetic operators and a multi-objective fitness function are designed for this representation. The approach has previously been illustrated using a binary class data set. In this paper we explore the possibility of using the algorithm on a multi-class data set. The accuracy of the rules produced by the evolutionary algorithm approach are compared to those obtained by a decision tree technique on the same data. The advantages of using an evolutionary classification technique over the more traditional decision tree structure are discussed.

GSOM sequence: An unsupervised dynamic approach for knowledge discovery in temporal data

A significant problem which arises during the process of knowledge discovery is dealing with data which have temporal dependencies. The attributes associated with temporal data need to be processed differently from non temporal attributes. A typical approach to address this issue is to view temporal data as an ordered sequence of events. In this work, we propose a novel dynamic unsupervised learning approach to discover patterns in temporal data. The new technique is based on the Growing Self-Organization Map (GSOM), which is a structure adapting version of the Self-Organizing Map (SOM). The SOM is widely used in knowledge discovery applications due to its unsupervised learning nature, ease of use and visualization capabilities. The GSOM further enhances the SOM with faster processing, more representative cluster formation and the ability to control map spread. This paper describes a significant extension to the GSOM enabling it to be used to for analyzing data with temporal sequences. The similarity between two time dependent sequences with unequal length is estimated using the Dynamic Time Warping (DTW) algorithm incorporated into the GSOM. Experiments were carried out to evaluate the performance and the validity of the proposed approach using an audio-visual data set. The results demonstrate that the novel “GSOM Sequence” algorithm improves the accuracy and validity of the clusters obtained.

Predicting bad credit risk: an evolutionary approach

This paper considers classification of binary valued data with unequal misclassification costs. This is a pertinent consideration in many applications of data mining, specifically in the area of credit scoring. An evolutionary algorithm is introduced and employed to generate rule systems for classification. In addition to the misclassification costs various other properties of the classification systems generated by the evolutionary algorithm, such as accuracy and coverage, are considered and discussed.

A Probabilistic Neural Network Approach to Modeling the Impact of Tobacco Control Policies by Gender

This paper evaluates the impact of tobacco control policies on female and male smokers. The data used in this study are from the 2002-2006 International Tobacco Control Four Country Survey. Based on eleven smokers’ motivational attributes used in the survey, principle component analysis is used to identify tobacco control policy drivers which are labeled as personal concerns, cigarette price, environmental restrictions and social encouragement. To examine the relative impact degrees of these four policy drivers on the groups of female and male smokers for their quit attempts, probabilistic neural network models are developed using hypothetical policy impacted populations. The experimental result shows that the most significant motivator for female smokers to make a quit attempt is their personal concerns. For male smokers, social encouragement plays a dominant role for them to make a quit attempt. The result indicates that smoking restrictions in public places or at workplace can somewhat encourage them to make a quit attempt. However, increasing the cigarette price is less likely to affect the MQA rate of both female and male smokers.

Innovation in Tobacco Control: Managing a Global Network with No Central Control

This paper develops a new analysis framework to examine the complex interactions within a tobacco control system in relation to the effects of tobacco control instruments. To develop the framework, we critically review the current status of tobacco control, including what instruments have been put into place, how much these instruments are helping, what problems we still have, and the reasons for current problems. The framework presents the architecture of a tobacco control system and its dynamic cycle of policy making, enactment, monitoring and refinement for analyzing tobacco control issues. To help develop new effective tobacco control instruments, we propose a conceptual model and use smoke-free places as an example for illustrating the innovation process. The analysis framework and the conceptual model have the potential to help manage tobacco control policy innovation in the process of decision-making, decision-refinement and ongoing management of government activities.

Evolutionary Rule Generation and its Application to Credit Scoring

Credit scoring is an important financial application area, concerned with assessing the likely risk of customers defaulting on granted credit. These customers may be bank customers borrowing money, or retail customers being sold goods on a deferred payment scheme. In order to model the relationship between the characteristics of a customer (financial and demographic) and their measured credit risk, data is collected and analysed. The developed model can then be used with new customers to determine their expected behaviour in an effort to automate the decision making process. The task of learning to classify objects in a database according to their attributes or characteristics has been tackled by many approaches over the decades. Statistical approaches such as logistic regression and discriminant analysis have yielded over recent years to soft computing methods such as neural networks, fuzzy logic and genetic algorithms. One of the advantages of soft computing techniques lies in their modeling capabilities in the presence of noisy, imprecise, inaccurate, or missing data. These considerations become specially important in applications such as credit scoring where the data attributes are likely to be incomplete, the classification of the customers as good or bad credit risks may be erroneous, and the costs of misclassification are particularly high.

Simulating Pricing Behaviours Using a Genetic Algorithm

Retail petrol prices in Australia are monitored by the federal government which sets the base price for petrol that oil companies must follow. Even though current regulations prohibit the companies from colluding, some flexibility over the actual retail price of petrol is allowed. This paper examines the oligopolistic behaviour of the petrol sellers in the API (Australian Petroleum Industry) using game theory and a genetic algorithm (GA). Experiments based on the API retail marketplace interaction were conducted with particular consideration given to the API rebate system. The major oil companies may set their petrol price below a fixed target price, but if they do so, they must rebate their sellers with the difference. Initial results suggest that game theory concepts and GAs are suitable tools for studying the API. Further work related to this project includes incorporating more realistic constraints into the system, better representation of the data in the model as well as comparing the results with human experts.

A new iterative fuzzy clustering algorithm for multiple imputation of missing data

This paper proposes a new iterative fuzzy clustering (IFC) algorithm to impute missing values of datasets. The information provided by fuzzy clustering is used to update the imputed values through iterations. The performance of the IFC algorithm is examined by conducting experiments on three commonly used datasets and a case study on a city mobility database. Experimental results show that the IFC algorithm not only works well for datasets with a small number of missing values but also provides an effective imputation result for datasets where the proportion of missing data is high.