Emmanuel Sirimal Silva

A review of data mining applications in crime

Crime continues to remain a severe threat to all communities and nations across the globe alongside the sophistication in technology and processes that are being exploited to enable highly complex criminal activities. Data mining, the process of uncovering hidden information from Big Data, is now an important tool for investigating, curbing and preventing crime and is exploited by both private and government institutions around the world. The primary aim of this paper is to provide a concise review of the data mining applications in crime. To this end, the paper reviews over 100 applications of data mining in crime, covering a substantial quantity of research to date, presented in chronological order with an overview table of many important data mining applications in the crime domain as a reference directory. The data mining techniques themselves are briefly introduced to the reader and these include entity extraction, clustering, association rule mining, decision trees, support vector machines, naive Bayes rule, neural networks and social network analysis amongst others.

From nature to maths

Many scientific fields consider accurate and reliable forecasting methods as important decision-making tools in the modern age amidst increasing volatility and uncertainty. As such there exists an opportune demand for theoretical developments which can result in more accurate forecasts. Inspired by Colonial Theory, this paper seeks to bring about considerable improvements to the field of time series analysis and forecasting by identifying certain core characteristics of Colonial Theory which are subsequently exploited in introducing a novel approach for the grouping step of subspace based methods. The proposed algorithm shows promising results in terms of improved performances in noise filtering and forecasting of time series. The reliability and validity of the proposed algorithm is evaluated and compared with popular forecasting models with the results being thoroughly evaluated for statistical significance and thereby adding more confidence and value to the findings of this research.

Optimizing Bicoid Signal Extraction

Signal extraction and analysis is of great importance, not only in fields such as economics and meteorology, but also in genetics and even biomedicine. There exists a range of parametric and nonparametric techniques which can perform signal extractions. However, the aim of this paper is to define a new approach for optimising signal extraction from bicoid gene expression profile. Having studied both parametric and nonparametric signal extraction techniques, we identified the lack of specific criteria enabling users to select the optimal signal extraction parameters. Exploiting the expression profile of bicoid gene, which is a maternal segmentation coordinate gene found in Drosophila melanogaster, we introduce a new approach for optimising the signal extraction using a nonparametric technique. The underlying criteria are based on the distribution of the residual, more specifically its skewness.

Forecasting energy data with a time lag into the future and Google trends

This paper presents a new idea for a forecasting approach which seeks to exploit the information contained within US EIA energy forecasts and related Google trends data for generating a new and improved forecast. The novel forecasting approach can be exploited by using a multivariate system which can consider data with different series lengths and a time lag into the future. Using real historical data, an official forecast for the same variable, and Google Trends search data, we illustrate the possibility of generating a comparatively more accurate forecast for an energy-related variable. The accuracy of the newly generated forecasts are evaluated by comparing with the actual observations and the official forecast itself. We find that the novel forecasting idea cangenerate promising results which call for further in-depth research into developing and improving this multivariate forecasting approach.

Big Data: a big opportunity for the petroleum and petrochemical industry

The Petroleum and Petrochemical (P&P) industry is home to the most traded commodity in the world, i.e. oil. Recently, this industry has been struggling to make ends meet with top lines being affected by falling oil prices and bottom lines being squeezed further via increasing operational costs. It is against this backdrop that this paper seeks to identify and summarise the positive influence that the adoption of Big Data can have on the P&P industry. Exhaustive research is carried out on the industry’s engagement and adoption of Big Data in upstream, midstream and downstream operations to concisely summarise the varied applications and the potential benefits. Our research indicates that the upstream sector is actively engaging with Big Data to achieve efficiency gains while the midstream and downstream sectors are lagging behind. Overall, it is evident that the P&P industry can find solutions to its aching financial and productivity issues by embracing of Big Data.

Estimation of protein diffusion parameters: Estimation of protein diffusion parameters

Protein diffusion offers an essential and elegant mechanism for morphogen gradient formation. Morphogens are signalling molecules that emanate from a particular region of the cell and create a gradient which has an impact on most biological processes, cell signalling and embryonic development. Using a method that is based on Singular Spectrum Analysis, we estimate parameters introduced in the Synthesis Diffusion Degradation model which is a commonly applied model for a transcription factor known as Bicoid. Our findings, consistent with simulation results, indicate that the proposed method can be practically applied as an enhanced parameter estimation technique with reduced sensitivity to various levels of noise.

Vector and recurrent singular spectrum analysis: which is better at forecasting?

ABSTRACT Singular spectrum analysis (SSA) is an increasingly popular and widely adopted filtering and forecasting technique which is currently exploited in a variety of fields. Given its increasing application and superior performance in comparison to other methods, it is pertinent to study and distinguish between the two forecasting variations of SSA. These are referred to as Vector SSA (SSA-V) and Recurrent SSA (SSA-R). The general notion is that SSA-V is more robust and provides better forecasts than SSA-R. This is especially true when faced with time series which are non-stationary and asymmetric, or affected by unit root problems, outliers or structural breaks. However, currently there exists no empirical evidence for proving the above notions or suggesting that SSA-V is better than SSA-R. In this paper, we evaluate out-of-sample forecasting capabilities of the optimised SSA-V and SSA-R forecasting algorithms via a simulation study and an application to 100 real data sets with varying structures, to provide a statistically reliable answer to the question of which SSA algorithm is best for forecasting at both short and long run horizons based on several important criteria.