Daswin De Silva

A Data Mining Framework for Electricity Consumption Analysis From Meter Data

This paper presents a novel data mining framework for the exploration and extraction of actionable knowledge from data generated by electricity meters. Although a rich source of information for energy consumption analysis, electricity meters produce a voluminous, fast-paced, transient stream of data that conventional approaches are unable to address entirely. In order to overcome these issues, it is important for a data mining framework to incorporate functionality for interim summarization and incremental analysis using intelligent techniques. The proposed Incremental Summarization and Pattern Characterization (ISPC) framework demonstrates this capability. Stream data is structured in a data warehouse based on key dimensions enabling rapid interim summarization. Independently, the IPCL algorithm incrementally characterizes patterns in stream data and correlates these across time. Eventually, characterized patterns are consolidated with interim summarization to facilitate an overall analysis and prediction of energy consumption trends. Results of experiments conducted using the actual data from electricity meters confirm applicability of the ISPC framework.

Incremental pattern characterization learning and forecasting for electricity consumption using smart meters

This paper presents a novel methodology for the incremental characterization and prediction of electricity consumption based on smart meter readings. A self-learning algorithm is developed to incrementally discover patterns in a data stream environment and sustain acquired knowledge for subsequent learning. It generates an evolving columnar structure composed of learning outcomes from each phase. This columnar structure characterizes electricity consumption and thus exposes significant patterns and continuity over time. The proposed technique is applied to smart meter data collected from RMIT University premises. Results show the potential for incremental pattern characterization learning in electricity consumption analysis and forecasting.

Semi-supervised classification of characterized patterns for demand forecasting using smart electricity meters

Smart meters are being gradually adopted by energy providers for commercial use due to multiple benefits. The extraction of actionable knowledge from smart meter readings can lead to informed decision-making in demand forecasting and consumption analysis. This paper extends an incremental learning approach for pattern characterization in a smart meter data stream environment, with the incorporation of a semi-supervised classification feature. The incremental pattern characterization learning (IPCL) algorithm autonomously learns from a smart meter environment and accumulates patterns in a columnar structure. The introduction of semi-supervised classification improves the quality and usability of the learning outcomes. We report outcomes demonstrating the classification of incremental learning outcomes, separation of cyclic patterns from exceptions, and a capacity to interpose new dimensions from the problem domain.

Incremental knowledge acquisition and self learning from text

Incremental learning is a core necessity in developments towards intelligent machines. Artificial learning as implemented in contemporary neural network algorithms does not fully encompass an incremental, autonomous learning capacity. In this paper we present a self learning algorithm capable of incrementally acquiring knowledge across learning periods. A dynamic unsupervised learning algorithm, the GSOM algorithm, forms the basis of the presented incrementally knowledge acquiring self learning (IKASL) algorithm, to which we have introduced a layer of aggregation for continuous learning, knowledge acquisition and retention. We also present a novel application of the IKASL algorithm for continuous learning of hidden patterns from semantics of text.

Trends in penile cancer: a comparative study between Australia, England and Wales, and the US

PurposeTo investigate and compare the trends in incidence and mortality of penile cancer between Australia, England and Wales, and the US, and provide hypotheses for these trends.MethodsCancer registry data from 1982 to 2005 inclusive were obtained from Australia, England and Wales, and the United States. From these data, age-specific, -standardised and mortality:incidence ratios were calculated, and compared.ResultsThe overall incidence of penile cancer in England and Wales (1.44 per 100,000 man-years) was higher than in Australia (0.80 per 100,000), and the US (0.66 per 100,000). Incidence of penile cancer in all three countries has remained relatively stable over time. Similarly, although the mortality rates were also higher in England and Wales (0.37 per 100,000 man-years) compared to Australia (0.18 per 100,000) and the US (0.15 per 100,000), the mortality/incidence ratios were similar for all three countries.ConclusionsPenile cancer incidence is low, affecting mainly older men. Rates differ between the three countries, being twice as common in England and Wales as in the other studied regions. Circumcision rates have a potential influence on these rates but are not the sole explanation for the variation.

Incidence of bladder cancer in sri lanka: analysis of the cancer registry data and review of the incidence of bladder cancer in the South asian population.

Purpose To investigate the incidence of bladder cancer (BC) in Sri Lanka and to compare risk factors and outcomes with those of other South Asian nations and South Asian migrants to the United Kingdom (UK) and the United States (US). Materials and Methods The incidence of BC in Sri Lanka was examined by using two separate cancer registry databases over a 5-year period. Smoking rates were compiled by using a population-based survey from 2001 to 2009 and the relative risk was calculated by using published data. Results A total of 637 new cases of BC were diagnosed over the 5-year period. Sri Lankan BC incidence increased from 1985 but remained low (1.36 and 0.3 per 100,000 in males and females) and was similar to the incidence in other South Asian countries. The incidence was lower, however, than in migrant populations in the US and the UK. In densely populated districts of Sri Lanka, these rates almost doubled. Urothelial carcinoma accounted for 72%. The prevalence of male smokers in Sri Lanka was 39%, whereas Pakistan had higher smoking rates with a 6-fold increase in BC. Conclusions Sri Lankan BC incidence was low, similar to other South Asian countries (apart from Pakistan), but the actual incidence is likely higher than the cancer registry rates. Smoking is likely to be the main risk factor for BC. Possible under-reporting in rural areas could account for the low rates of BC in Sri Lanka. Any genetic or environmental protective effects of BC in South Asians seem to be lost on migration to the UK or the US and with higher levels of smoking, as seen in Pakistan.

Automatic event detection in microblogs using incremental machine learning

The global popularity of microblogs has led to an increasing accumulation of large volumes of text data on microblogging platforms such as Twitter. These corpora are untapped resources to understand social expressions on diverse subjects. Microblog analysis aims to unlock the value of such expressions by discovering insights and events of significance hidden among swathes of text. Besides velocity; diversity of content, brevity, absence of structure and time‐sensitivity are key challenges in microblog analysis. In this paper, we propose an unsupervised incremental machine learning and event detection technique to address these challenges. The proposed technique separates a microblog discussion into topics to address the key problem of diversity. It maintains a record of the evolution of each topic over time. Brevity, time‐sensitivity and unstructured nature are addressed by these individual topic pathways which contribute to generate a temporal, topic‐driven structure of a microblog discussion. The proposed event detection method continuously monitors these topic pathways using multiple domain‐independent event indicators for events of significance. The autonomous nature of topic separation, topic pathway generation, new topic identification and event detection, appropriates the proposed technique for extensive applications in microblog analysis. We demonstrate these capabilities on tweets containing #microsoft and tweets containing #obama.

A novel framework for automated, intelligent extraction and analysis of online support group discussions for cancer related outcomes

Cancer care requires an extensive network of support for each patient, from the first discussion of treatment options, subsequent treatment and post-treatment care to surveillance. These increasing expectations of cancer care are a formidable challenge to healthcare systems across the world. This elemental issue arises when allocating appropriate resources to address specific patient needs and expectations.

An Intelligent Content Discovery Technique for Health Portal Content Management

Background Continuous content management of health information portals is a feature vital for its sustainability and widespread acceptance. Knowledge and experience of a domain expert is essential for content management in the health domain. The rate of generation of online health resources is exponential and thereby manual examination for relevance to a specific topic and audience is a formidable challenge for domain experts. Intelligent content discovery for effective content management is a less researched topic. An existing expert-endorsed content repository can provide the necessary leverage to automatically identify relevant resources and evaluate qualitative metrics. Objective This paper reports on the design research towards an intelligent technique for automated content discovery and ranking for health information portals. The proposed technique aims to improve efficiency of the current mostly manual process of portal content management by utilising an existing expert-endorsed content repository as a supporting base and a benchmark to evaluate the suitability of new content Methods A model for content management was established based on a field study of potential users. The proposed technique is integral to this content management model and executes in several phases (ie, query construction, content search, text analytics and fuzzy multi-criteria ranking). The construction of multi-dimensional search queries with input from Wordnet, the use of multi-word and single-word terms as representative semantics for text analytics and the use of fuzzy multi-criteria ranking for subjective evaluation of quality metrics are original contributions reported in this paper. Results The feasibility of the proposed technique was examined with experiments conducted on an actual health information portal, the BCKOnline portal. Both intermediary and final results generated by the technique are presented in the paper and these help to establish benefits of the technique and its contribution towards effective content management. Conclusions The prevalence of large numbers of online health resources is a key obstacle for domain experts involved in content management of health information portals and websites. The proposed technique has proven successful at search and identification of resources and the measurement of their relevance. It can be used to support the domain expert in content management and thereby ensure the health portal is up-to-date and current.

Multivariate Data-Driven Decision Guidance for clinical scientists

Clinical decision-support is gaining widespread attention as medical institutions and governing bodies turn towards utilising better information management for effective and efficient healthcare delivery and quality assured outcomes. A mass of data across all stages, from disease diagnosis to palliative care, is further indication of the opportunities and challenges created for effective data management, analysis, prediction and optimization techniques as parts of knowledge management in clinical environments. A Data-driven Decision Guidance Management System (DD-DGMS) architecture can encompass solutions into a single closed-loop integrated platform to empower clinical scientists to seamlessly explore a multivariate data space in search of novel patterns and correlations to inform their research and practice. The paper describes the components of such an architecture, which includes a robust data warehouse as an infrastructure for comprehensive clinical knowledge management. The proposed DD-DGMS architecture incorporates the dynamic dimensional data model as its elemental core. Given the heterogeneous nature of clinical contexts and corresponding data, the dimensional data model presents itself as an adaptive model that facilitates knowledge discovery, distribution and application, which is essential for clinical decision support. The paper reports on a trial of the DD-DGMS system prototype conducted on diabetes screening data which further establishes the relevance of the proposed architecture to a clinical context.