Description:
Objective: To compare Orchard Sports Injury Classification System (OSICS-10) sports medicine diagnoses assigned by a clinical and non-clinical coder. Design: Assessment of intercoder agreement. Setting: Community Australian football. Participants: 1082 standardised injury surveillance records. Main outcome measurements: Direct comparison of the four-character hierarchical OSICS-10 codes assigned by two independent coders (a sports physician and an epidemiologist). Adjudication by a third coder (biomechanist). Results: The coders agreed on the first character 95% of the time and on the first two characters 86% of the time. They assigned the same four-digit OSICS-10 code for only 46% of the 1082 injuries. The majority of disagreements occurred for the third character; 85% were because one coder assigned a non-specific 'X' code. The sports physician code was deemed correct in 53% of cases and the epidemiologist in 44%. Reasons for disagreement included the physician not using all of the collected information and the epidemiologist lacking specific anatomical knowledge. Conclusions: Sports injury research requires accurate identification and classification of specific injuries and this study found an overall high level of agreement in coding according to OSICS-10. The fact that the majority of the disagreements occurred for the third OSICS character highlights the fact that increasing complexity and diagnostic specificity in injury coding can result in a loss of reliability and demands a high level of anatomical knowledge. Injury report form details need to reflect this level of complexity and data management teams need to include a broad range of expertise. Copyright Article author (or their employer) 2012.
Description:
Objective: To compare Orchard Sports Injury Classification System (OSICS-10) sports medicine diagnoses assigned by a clinical and non-clinical coder. Design: Assessment of intercoder agreement. Setting: Community Australian football. Participants: 1082 standardised injury surveillance records. Main outcome measurements: Direct comparison of the four-character hierarchical OSICS-10 codes assigned by two independent coders (a sports physician and an epidemiologist). Adjudication by a third coder (biomechanist). Results: The coders agreed on the first character 95% of the time and on the first two characters 86% of the time. They assigned the same four-digit OSICS-10 code for only 46% of the 1082 injuries. The majority of disagreements occurred for the third character; 85% were because one coder assigned a non-specific 'X' code. The sports physician code was deemed correct in 53% of cases and the epidemiologist in 44%. Reasons for disagreement included the physician not using all of the collected information and the epidemiologist lacking specific anatomical knowledge. Conclusions: Sports injury research requires accurate identification and classification of specific injuries and this study found an overall high level of agreement in coding according to OSICS-10. The fact that the majority of the disagreements occurred for the third OSICS character highlights the fact that increasing complexity and diagnostic specificity in injury coding can result in a loss of reliability and demands a high level of anatomical knowledge. Injury report form details need to reflect this level of complexity and data management teams need to include a broad range of expertise. Copyright Article author (or their employer) 2012.
Description:
Text Classification is the task of mapping a document into one or more classes based on the presence or absence of words (or features) in the document. It is intensively being studied and different classification techniques and algorithms have been developed. This thesis focuses on classification of online documents that has become more critical with the development of World Wide Web. The WWW vastly increases the availability of on-line documents in digital format and has highlighted the need to classify them. From this background, we have noted the emergence of “automatic Web Classification”. These mainly concentrate on classifying HTML-like documents into classes or categories by not only using the methods that are inherited from the traditional Text Classification process, but also utilizing the extra information provided only by Web pages. Our work is based on the fact that, Web documents, contain not only ordinary features (words) but also extra information, such as meta-data and hyperlinks that can be used to advantage the classification process. The aim of this research is to study various ways of using the extra information, in particularly, hyperlink information provided by HTML-documents (Web pages). The merit of the approach, developed in this thesis, is its simplicity, compared with existing approaches. We present different approaches of using hyperlink information to improve the effectiveness of web classification. Unlike other work in this area, we will only use the mappings between linked documents and their own class or classes. In this case, we only need to add a few features called linked-class features into the datasets, and then apply classifiers on them for classification. In the numerical experiments we adopted two wellknown Text Classification algorithms, Support Vector Machines and BoosTexter. The results obtained show that classification accuracy can be improved by using mixtures of ordinary and linked-class features. Moreover, out-links usually work better than in-links in classification. We also analyse and discuss the reasons behind this improvement.
Description:
Text Classification is the task of mapping a document into one or more classes based on the presence or absence of words (or features) in the document. It is intensively being studied and different classification techniques and algorithms have been developed. This thesis focuses on classification of online documents that has become more critical with the development of World Wide Web. The WWW vastly increases the availability of on-line documents in digital format and has highlighted the need to classify them. From this background, we have noted the emergence of “automatic Web Classification”. These mainly concentrate on classifying HTML-like documents into classes or categories by not only using the methods that are inherited from the traditional Text Classification process, but also utilizing the extra information provided only by Web pages. Our work is based on the fact that, Web documents, contain not only ordinary features (words) but also extra information, such as meta-data and hyperlinks that can be used to advantage the classification process. The aim of this research is to study various ways of using the extra information, in particularly, hyperlink information provided by HTML-documents (Web pages). The merit of the approach, developed in this thesis, is its simplicity, compared with existing approaches. We present different approaches of using hyperlink information to improve the effectiveness of web classification. Unlike other work in this area, we will only use the mappings between linked documents and their own class or classes. In this case, we only need to add a few features called linked-class features into the datasets, and then apply classifiers on them for classification. In the numerical experiments we adopted two wellknown Text Classification algorithms, Support Vector Machines and BoosTexter. The results obtained show that classification accuracy can be improved by using mixtures of ordinary and linked-class features. Moreover, out-links usually work better than in-links in classification. We also analyse and discuss the reasons behind this improvement.
Description:
Diagnostic radiology struggles to maintain high interpretation accuracy. Retrieval of past similar cases would help the inexperienced radiologist in the interpretation process. Character n-gram model has been effective in text retrieval context in languages such as Chinese where there are no clear word boundaries. We propose the use of visual character n-gram model for representation of image for classification and retrieval purposes. Regions of interests in mammographic images are represented with the character n-gram features. These features are then used as input to back-propagation neural network for classification of regions into normal and abnormal categories. Experiments on miniMIAS database show that character n-gram features are useful in classifying the regions into normal and abnormal categories. Promising classification accuracies are observed (83.33%) for fatty background tissue warranting further investigation. We argue that Classifying regions of interests would reduce the number of comparisons necessary for finding similar images from the database and hence would reduce the time required for retrieval of past similar cases.
Description:
Diagnostic radiology struggles to maintain high interpretation accuracy. Retrieval of past similar cases would help the inexperienced radiologist in the interpretation process. Character n-gram model has been effective in text retrieval context in languages such as Chinese where there are no clear word boundaries. We propose the use of visual character n-gram model for representation of image for classification and retrieval purposes. Regions of interests in mammographic images are represented with the character n-gram features. These features are then used as input to back-propagation neural network for classification of regions into normal and abnormal categories. Experiments on miniMIAS database show that character n-gram features are useful in classifying the regions into normal and abnormal categories. Promising classification accuracies are observed (83.33%) for fatty background tissue warranting further investigation. We argue that Classifying regions of interests would reduce the number of comparisons necessary for finding similar images from the database and hence would reduce the time required for retrieval of past similar cases.
Description:
Data mining is often performed with datasets associated with diseases in order to increase insights that can ultimately lead to improved prevention or treatment. Classification algorithms can achieve high levels of predictive accuracy but have limited application for facilitating the insight that leads to deeper understanding of aspects of the disease. This is because the representation of knowledge that arises from classification algorithms is too opaque, too complex or too sparse to facilitate insight. Clustering, association and visualisation approaches enable greater scope for clinicians to be engaged in a way that leads to insight, however predictive accuracy is compromised or non-existent. This research investigates the practical applications of Automated Weighted Sum, (AWSum), a classification algorithm that provides accuracy comparable to other techniques whilst providing some insight into the data. This is achieved by calculating a weight for each feature value that represents its influence on the class value. Clinicians are very familiar with weighted sum scoring scales so the internal representation is intuitive and easily understood. This paper presents results from the use of the AWSum approach with data from patients suffering from Cystic Fibrosis.
Description:
Data mining is often performed with datasets associated with diseases in order to increase insights that can ultimately lead to improved prevention or treatment. Classification algorithms can achieve high levels of predictive accuracy but have limited application for facilitating the insight that leads to deeper understanding of aspects of the disease. This is because the representation of knowledge that arises from classification algorithms is too opaque, too complex or too sparse to facilitate insight. Clustering, association and visualisation approaches enable greater scope for clinicians to be engaged in a way that leads to insight, however predictive accuracy is compromised or non-existent. This research investigates the practical applications of Automated Weighted Sum, (AWSum), a classification algorithm that provides accuracy comparable to other techniques whilst providing some insight into the data. This is achieved by calculating a weight for each feature value that represents its influence on the class value. Clinicians are very familiar with weighted sum scoring scales so the internal representation is intuitive and easily understood. This paper presents results from the use of the AWSum approach with data from patients suffering from Cystic Fibrosis.
Description:
The Australian Adverse Drug Reaction Advisory Committee (ADRAC) database has been collected and maintained by the Therapeutic Goods Administration. In this paper we study a part of his database (Card2) which contains records having just reactions from the Cardiovascular group. Drug-reaction relationships are presented by a vector of degrees which shows the degree of association of a drug with each class of reactions. In this work we examine these relationships in the classification of reaction outcomes. A modified version of the fuzzy derivative method (FDM2) is used for classification.
Description:
This article reports research conducted on students’ perceptions of assessment in science classes in Queensland and Western Australia. A specially developed instrument, the Student Perceptions of Assessment Questionnaire (SPAQ) which assesses Congruence with Planned Learning, Authenticity, Student Consultation, Transparency, and Diversity was used to collect data from 3,055 students. Hierarchical cluster analysis resulted in a four cluster solution being accepted. While one cluster of 799 students held positive perceptions of assessment, another cluster of 640 students held negative views. The SPAQ allows for a greater focus on classroom-based perceptions of assessment rather than crude external accountability measures that decontextualise classroom assessment.
Description:
Drensky and Lakatos (Lecture Notes in Computer Science, 357 (Springer, Berlin, 1989), pp. 181-188) have established a convenient property of certain ideals in polynomial quotient rings, which can now be used to determine error-correcting capabilities of combined multiple classifiers following a standard approach explained in the well-known monograph by Witten and Frank (Data Mining: Practical Machine Learning Tools and Techniques (Elsevier, Amsterdam, 2005)). We strengthen and generalise the result of Drensky and Lakatos by demonstrating that the corresponding nice property remains valid in a much larger variety of constructions and applies to more general types of ideals. Examples show that our theorems do not extend to larger classes of ring constructions and cannot be simplified or generalised.
Description:
Drensky and Lakatos (Lecture Notes in Computer Science, 357 (Springer, Berlin, 1989), pp. 181-188) have established a convenient property of certain ideals in polynomial quotient rings, which can now be used to determine error-correcting capabilities of combined multiple classifiers following a standard approach explained in the well-known monograph by Witten and Frank (Data Mining: Practical Machine Learning Tools and Techniques (Elsevier, Amsterdam, 2005)). We strengthen and generalise the result of Drensky and Lakatos by demonstrating that the corresponding nice property remains valid in a much larger variety of constructions and applies to more general types of ideals. Examples show that our theorems do not extend to larger classes of ring constructions and cannot be simplified or generalised.
Description:
Solar generation systems are globally extending in terms of scale and number, which highlights the increasing importance of solar power forecast. In this paper, a day-ahead solar power prediction method is proposed including 1) a novel feature selecting/clustering approach based on relevancy and redundancy criteria and 2) an innovative hybrid-classification-regression forecasting engine. The proposed feature selecting/clustering approach filters out irrelevant features and partitions relevant features to two separate subsets to decrease the redundancy of features. Each of these two subsets is separately trained by one forecasting engine and the final solar power prediction of the proposed method is obtained by a relevancy-based combination of these two forecasts. The proposed forecasting engine classifies the historical data based on the learnability of its constituent regression models and assigns each class of training samples to one regression model. Each regression model predicts the outputs of the test samples that belong to its class. The effectiveness of the proposed solar power prediction method is illustrated by testing on two real-world solar farms.
Description:
In this paper, we revisited the classical technique of Regularized Least Squares (RLS) for the classification of large-scale nonlinear data. Specifically, we focus on a low-rank formulation of RLS and show that it has linear time complexity in the data size only and does not rely on the number of labels and features for problems with moderate feature dimension. This makes low-rank RLS particularly suitable for classification with large data sets. Moreover, we have proposed a general theorem for the closed-form solutions to the Leave-One-Out Cross Validation (LOOCV) estimation problem in empirical risk minimization which encompasses all types of RLS classifiers as special cases. This eliminates the reliance on cross validation, a computationally expensive process for parameter selection, and greatly accelerate the training process of RLS classifiers. Experimental results on real and synthetic large-scale benchmark data sets have shown that low-rank RLS achieves comparable classification performance while being much more efficient than standard kernel SVM for nonlinear classification. The improvement in efficiency is more evident for data sets with higher dimensions.
Description:
This article is devoted to experimental investigation of classification algorithms for analysis of ITS dataset. We introduce and consider a novel k-committees alogorithm for classification and compare it with the discrete k- means and nearest neighbour algorithms. The ITS dataset consists of nuclear ribosomal DNA sequences, where rather sophisticated alignment scores have to be used as a measure of distance. These scores do not form Minkowski metric and the sequences cannot be regarded as points in a finite dimensional space. This is why it is necessary to develop novel algorithms and adjust familiar ones. We present the results of experiments comparing the efficiency of three classification methods in their ability to achieve agreement with classes published in the biological literature before. It turns out that our algorithms are efficient and can be used to obtain biologically significant classifications. A simplified version of a synthetic dataset, where the k-committees classifier out performs k-means and Nearest Neighbour classifiers, is also presented.
Description:
Missing values may be present in data without undermining its use for diagnostic / classification purposes but compromise application of readily available software. Surrogate entries can remedy the situation, although the outcome is generally unknown. Discretization of continuous attributes renders all data nominal and is helpful in dealing with missing values; particularly, no special handling is required for different attribute types. A number of classifiers exist or can be reformulated for this representation. Some classifiers can be reinvented as data completion methods. In this work the Decision Tree, Nearest Neighbour, and Naive Bayesian methods are demonstrated to have the required aptness. An approach is implemented whereby the entered missing values are not necessarily a close match of the true data; however, they intend to cause the least hindrance for classification. The proposed techniques find their application particularly in medical diagnostics. Where clinical data represents a number of related conditions, taking Cartesian product of class values of the underlying sub-problems allows narrowing down of the selection of missing value substitutes. Real-world data examples, some publically available, are enlisted for testing. The proposed and benchmark methods are compared by classifying the data before and after missing value imputation, indicating a significant improvement.
Description:
Missing values may be present in data without undermining its use for diagnostic / classification purposes but compromise application of readily available software. Surrogate entries can remedy the situation, although the outcome is generally unknown. Discretization of continuous attributes renders all data nominal and is helpful in dealing with missing values; particularly, no special handling is required for different attribute types. A number of classifiers exist or can be reformulated for this representation. Some classifiers can be reinvented as data completion methods. In this work the Decision Tree, Nearest Neighbour, and Naive Bayesian methods are demonstrated to have the required aptness. An approach is implemented whereby the entered missing values are not necessarily a close match of the true data; however, they intend to cause the least hindrance for classification. The proposed techniques find their application particularly in medical diagnostics. Where clinical data represents a number of related conditions, taking Cartesian product of class values of the underlying sub-problems allows narrowing down of the selection of missing value substitutes. Real-world data examples, some publically available, are enlisted for testing. The proposed and benchmark methods are compared by classifying the data before and after missing value imputation, indicating a significant improvement.
Description:
This article presents experimental results devoted to a new application of the novel clustering technique introduced by the authors recently. Our aim is to facilitate the application of robust and stable consensus functions in information security, where it is often necessary to process large data sets and monitor outcomes in real time, as it is required, for example, for intrusion detection. Here we concentrate on the particular case of application to profiling of phishing websites. First, we apply several independent clustering algorithms to a randomized sample of data to obtain independent initial clusterings. Silhouette index is used to determine the number of clusters. Second, we use a consensus function to combine these independent clusterings into one consensus clustering . Feature ranking is used to select a subset of features for the consensus function. Third, we train fast supervised classification algorithms on the resulting consensus clustering in order to enable them to process the whole large data set as well as new data. The precision and recall of classifiers at the final stage of this scheme are critical for effectiveness of the whole procedure. We investigated various combinations of three consensus functions, Cluster-Based Graph Formulation (CBGF), Hybrid Bipartite Graph Formulation (HBGF), and Instance-Based Graph Formulation (IBGF) and a variety of supervised classification algorithms. The best precision and recall have been obtained by the combination of the HBGF consensus function and the SMO classifier with the polynomial kernel.