Journal of Advanced Computational Intelligence and Intelligent Informatics, Mar 20, 2000
In the k-Nearest Neighbour (kNN) algorithm, the classification of a new sample is determined by t... more In the k-Nearest Neighbour (kNN) algorithm, the classification of a new sample is determined by the class of its k nearest neighbours. The performance of the kNN algorithm is influenced by three main factors: (1) the distance metric used to locate the nearest neighbours; (2) the decision rule used to derive a classification from the k-nearest neighbours; and (3) the number of neighbours used to classify the new sample. Using k = 1, 3, or 5 nearest neighbours, this study uses a Genetic Algorithm (GA) to find the optimal non-Euclidean distance metric in the kNN algorithm and examines two alternative methods (Majority Rule and Bayes Rule) to derive a classification from the k nearest neighbours. This modified algorithm was evaluated on two real-world medical fraud problems. The General Practitioner (GP) database is a 2-class problem in which GPs are classified as either practising appropriately or inappropriately. The ’.Doctor-Shoppers’ database is a 5-class problem in which patients are classified according to the likelihood that they are ’doctor-shoppers’. Doctor-shoppers are patients who consult many physicians in order to obtain multiple prescriptions of drugs of addiction in excess of their own therapeutic need. In both applications, classification accuracy was improved by optimising the distance metric in the kNN algorithm. The agreement rate on the GP dataset improved from around 70% (using Euclidean distance) to 78 % (using an optimised distance metric), and from about 55% to 82% on the Doctor Shopper’s dataset. Differences in either the decision rule or the number of nearest neighbours had little or no impact on the classification performance of the kNN algorithm. The excellent performance of the kNN algorithm when the distance metric is optimised using a genetic algorithm paves the way for its application in the real world fraud detection problems faced by the Health Insurance Commission (HIC).
This paper discusses three issues relevant to leadership personality research. The first compares... more This paper discusses three issues relevant to leadership personality research. The first compares interindividual to intraindividual research strategies. The former explains differences between individuals such as leaders and nonleaders and the latter explains differences within leaders. It is suggested that intraindividual strategies are better than interindividual strategies for demonstrating the role personality characteristics play in leadership. The second discusses situationally determined leadership measurements. It is recommended that researchers should use leadership measurements which are related to work behaviour rather than use general personality trait measurements. The third reviews the relationship between personality variables and leadership behaviours. It is pointed out that leadership research has concentrated on identifying the personality characteristics of leaders and the behaviours they use but few attempts have been made to link the two sets of variables.
The application of data mining and knowledge discovery techniques to medical and health datasets ... more The application of data mining and knowledge discovery techniques to medical and health datasets is a rewarding but highly challenging area. Not only are the datasets large, complex, heterogeneous, hierarchical, time-varying and of varying quality but there exists asubstantial medical knowledge base which demands a robust collaboration between the data miner and the health professional(s) if useful information is to be extracted.This paper presents the experiences of the authors and others in applying exploratory data mining techniques to medical, health and clinical data. In so doing, it elicits a number of general issues and provides pointers to possible areas of future research in data mining and knowledge discovery more broadly.
This paper describes a colour approach to doing data processing and computations and it illustrat... more This paper describes a colour approach to doing data processing and computations and it illustrates this with everyday applications. Everything can be represented as light with examples including text, numbers, objects and patterns. They can be each given a unique colour code. Computers that use light to perform data processing and computations are referred to either as ‘chromatic computers’ or ‘photonic computers’. Light can be decomposed into a colour spectrum and colours can be used to represent data and to do mathematical operations. Chromathics is a process of doing calculations and transformations with light. Light-based operations can be performed at speeds faster than digital ones and can compress data better than can be done with digital computers. This paper will explain both chromatic computers and chromathics. It will provide everyday examples to illustrate the use of a colour approach to computing.
In this paper the results of capstone analysis is applied to predict a weather outcome using a de... more In this paper the results of capstone analysis is applied to predict a weather outcome using a decision-tree model. It examines weather data of the capital cities of Australia in a 12 month period to see if the decision-tree models can predict rain in Sydney the next day. It produces a decision-tree model with the raw weather data for each capital city to make the predictions about this outcome. It also aggregates the raw data to provide a combined-city dataset. Finally, it combines and compresses the raw data for each city using capstone modelling. The capstone data for the cities is used to train another decision-tree model to see if this provides better predictions of rain in Sydney than those obtained from using single-city models and the combined-city model. The results of these comparisons and details of how capstoning works are provided in the paper
for individual, company, and other types of tax returns and to prepare business activity statemen... more for individual, company, and other types of tax returns and to prepare business activity statements for business taxes, including goods and services taxes or GST. Unfortunately, some agents abuse their positions of trust to defraud the tax system. One way they do this is by inflating the business deductions of their clients. The Australian Taxation Office (ATO) is responsible for identifying high-risk tax agents who are engaging in unacceptable practice. The methods described in this paper were aimed at identifying high-risk agents. In this paper, we report some results from modelling tax agent behavior using a distance-from-the-centroid (DFC) method with assistance from a genetic algorithm (GA). DFC is an example of what are called “instance-based learning methods.” These use known high-risk cases, or instances, to see if other cases have practice profiles that are similar to them. DFC works simply by identifying the center of gravity or centroid of a collection of known high-risk ...
2011 IEEE 11th International Conference on Data Mining Workshops, Dec 1, 2011
" workshop aims to provide a premier forum for sharing findings, knowledge, insight, experience a... more " workshop aims to provide a premier forum for sharing findings, knowledge, insight, experience and lessons in tackling potential challenges in discovering actionable knowledge from complex domain problems, promoting interaction and filling the gap between academia and business, and driving a paradigm shift from interesting hidden pattern mining to actionable knowledge discovery in varying data mining domains.
Journal of Advanced Computational Intelligence and Intelligent Informatics, Mar 20, 2000
In the k-Nearest Neighbour (kNN) algorithm, the classification of a new sample is determined by t... more In the k-Nearest Neighbour (kNN) algorithm, the classification of a new sample is determined by the class of its k nearest neighbours. The performance of the kNN algorithm is influenced by three main factors: (1) the distance metric used to locate the nearest neighbours; (2) the decision rule used to derive a classification from the k-nearest neighbours; and (3) the number of neighbours used to classify the new sample. Using k = 1, 3, or 5 nearest neighbours, this study uses a Genetic Algorithm (GA) to find the optimal non-Euclidean distance metric in the kNN algorithm and examines two alternative methods (Majority Rule and Bayes Rule) to derive a classification from the k nearest neighbours. This modified algorithm was evaluated on two real-world medical fraud problems. The General Practitioner (GP) database is a 2-class problem in which GPs are classified as either practising appropriately or inappropriately. The ’.Doctor-Shoppers’ database is a 5-class problem in which patients are classified according to the likelihood that they are ’doctor-shoppers’. Doctor-shoppers are patients who consult many physicians in order to obtain multiple prescriptions of drugs of addiction in excess of their own therapeutic need. In both applications, classification accuracy was improved by optimising the distance metric in the kNN algorithm. The agreement rate on the GP dataset improved from around 70% (using Euclidean distance) to 78 % (using an optimised distance metric), and from about 55% to 82% on the Doctor Shopper’s dataset. Differences in either the decision rule or the number of nearest neighbours had little or no impact on the classification performance of the kNN algorithm. The excellent performance of the kNN algorithm when the distance metric is optimised using a genetic algorithm paves the way for its application in the real world fraud detection problems faced by the Health Insurance Commission (HIC).
This paper discusses three issues relevant to leadership personality research. The first compares... more This paper discusses three issues relevant to leadership personality research. The first compares interindividual to intraindividual research strategies. The former explains differences between individuals such as leaders and nonleaders and the latter explains differences within leaders. It is suggested that intraindividual strategies are better than interindividual strategies for demonstrating the role personality characteristics play in leadership. The second discusses situationally determined leadership measurements. It is recommended that researchers should use leadership measurements which are related to work behaviour rather than use general personality trait measurements. The third reviews the relationship between personality variables and leadership behaviours. It is pointed out that leadership research has concentrated on identifying the personality characteristics of leaders and the behaviours they use but few attempts have been made to link the two sets of variables.
The application of data mining and knowledge discovery techniques to medical and health datasets ... more The application of data mining and knowledge discovery techniques to medical and health datasets is a rewarding but highly challenging area. Not only are the datasets large, complex, heterogeneous, hierarchical, time-varying and of varying quality but there exists asubstantial medical knowledge base which demands a robust collaboration between the data miner and the health professional(s) if useful information is to be extracted.This paper presents the experiences of the authors and others in applying exploratory data mining techniques to medical, health and clinical data. In so doing, it elicits a number of general issues and provides pointers to possible areas of future research in data mining and knowledge discovery more broadly.
This paper describes a colour approach to doing data processing and computations and it illustrat... more This paper describes a colour approach to doing data processing and computations and it illustrates this with everyday applications. Everything can be represented as light with examples including text, numbers, objects and patterns. They can be each given a unique colour code. Computers that use light to perform data processing and computations are referred to either as ‘chromatic computers’ or ‘photonic computers’. Light can be decomposed into a colour spectrum and colours can be used to represent data and to do mathematical operations. Chromathics is a process of doing calculations and transformations with light. Light-based operations can be performed at speeds faster than digital ones and can compress data better than can be done with digital computers. This paper will explain both chromatic computers and chromathics. It will provide everyday examples to illustrate the use of a colour approach to computing.
In this paper the results of capstone analysis is applied to predict a weather outcome using a de... more In this paper the results of capstone analysis is applied to predict a weather outcome using a decision-tree model. It examines weather data of the capital cities of Australia in a 12 month period to see if the decision-tree models can predict rain in Sydney the next day. It produces a decision-tree model with the raw weather data for each capital city to make the predictions about this outcome. It also aggregates the raw data to provide a combined-city dataset. Finally, it combines and compresses the raw data for each city using capstone modelling. The capstone data for the cities is used to train another decision-tree model to see if this provides better predictions of rain in Sydney than those obtained from using single-city models and the combined-city model. The results of these comparisons and details of how capstoning works are provided in the paper
for individual, company, and other types of tax returns and to prepare business activity statemen... more for individual, company, and other types of tax returns and to prepare business activity statements for business taxes, including goods and services taxes or GST. Unfortunately, some agents abuse their positions of trust to defraud the tax system. One way they do this is by inflating the business deductions of their clients. The Australian Taxation Office (ATO) is responsible for identifying high-risk tax agents who are engaging in unacceptable practice. The methods described in this paper were aimed at identifying high-risk agents. In this paper, we report some results from modelling tax agent behavior using a distance-from-the-centroid (DFC) method with assistance from a genetic algorithm (GA). DFC is an example of what are called “instance-based learning methods.” These use known high-risk cases, or instances, to see if other cases have practice profiles that are similar to them. DFC works simply by identifying the center of gravity or centroid of a collection of known high-risk ...
2011 IEEE 11th International Conference on Data Mining Workshops, Dec 1, 2011
" workshop aims to provide a premier forum for sharing findings, knowledge, insight, experience a... more " workshop aims to provide a premier forum for sharing findings, knowledge, insight, experience and lessons in tackling potential challenges in discovering actionable knowledge from complex domain problems, promoting interaction and filling the gap between academia and business, and driving a paradigm shift from interesting hidden pattern mining to actionable knowledge discovery in varying data mining domains.
Uploads
Papers by Warwick Graco