Shelton Peiris | The University of Sydney (original) (raw)

Papers by Shelton Peiris

Mathematics, Nov 21, 2022

This article is an open access article distributed under the terms and conditions of the Creative... more This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY

Factor analysis is an important multivariate statistical technique, which examines the structure ... more Factor analysis is an important multivariate statistical technique, which examines the structure of the interrelationships among a large number of variables, by determining a set of underlying dimensions or factors. This paper examines two examples of the use of factor analysis in business research. In the first application, factor analysis was utilised to identify the key strategy dimensions underlying 66 elements of new product development, identified in normative strategy literature, previous research into the determinants of new product success and failure, and previous studies of firms’ new product strategies. Cluster analysis was then used on the factor scores of identified dimensions to determine the different types of strategies that firms elect for their new product programs. The second example describes how factor analysis was employed to explore groupings of 83 indicators of destination competitiveness found in the tourism literature and the wider management literature. The results of the factor analysis were compared with a theoretical model of tourism competitiveness. Factor analysis seems to confirm many of the considerations identified by tourism practitioners and researchers, in a formal sense

The main objective of the study is to construct mathematical models using data from 1960 to 2000,... more The main objective of the study is to construct mathematical models using data from 1960 to 2000, collected by Australian Bureau of Agricultural and Resource Economics (ABARE), to predict yield of wheat in Australia, under varying assumptions. We have considered two approaches, namely a deterministic approach and a stochastic approach in constructing mathematical models for the yield of Australian wheat/grain yield for prediction purposes. Under the deterministic approach we developed a modified quadratic logistic model and some simulations were done using S-Plus. As an alternative, we introduced a random component, to take into account the large differences in the approaches taken to wheat and grain production in Australia. In applying this stochastic approach, we used a log-normal diffusion process

This chapter reports a new class of algorithms for the estimation of parameters using massive dat... more This chapter reports a new class of algorithms for the estimation of parameters using massive datasets. Certain properties of sub-totaling and repeated estimation of population parameters are used to establish a new statistical method for estimating summary characteristics of populations, and relationships between variables with extremely large datasets. The technique could be utilzed in distributed computing especially for applications in data mining, analysis and modeling with massive data sets

Series of critical dry spell lengths in 56 years in Anamaduwa are analysed to predict the length ... more Series of critical dry spell lengths in 56 years in Anamaduwa are analysed to predict the length of critical dry spells Both linear and nonlinear lime series approaches are tried to identify the best Jilted model By comparing various statistical indicators, bilinear model with auto regressive errors of order four is found to be the best model lo lit die critical dry spell lengths

Journal of Statistical Computation and Simulation

2021 International Conference on Data Science, Artificial Intelligence, and Business Analytics (DATABIA), 2021

From our previous study, we have known that only a small number of literatures have studied peatl... more From our previous study, we have known that only a small number of literatures have studied peatlands fire modeling in Indonesia. It is including our recent study on the prediction of the forest fire occurrence in the peatlands area using some machine learning classification techniques. In the previous empirical study using data from South Kalimantan Province in Indonesia, we found that the datasets are unbalanced between the two classes of data, i.e., the occurrence of fire hotspots and the nonoccurrence of fire hotspots areas. In this paper, the performance of the classification method is improved, by balancing the data using what so called Synthetic Minority Over-sampling Technique (SMOTE). In the empirical results, we show the performance of the classification results on the balanced data are mixed. It is found that only using the ensemble AdaBoost with SMOTE balanced data the performance of the methods has always been improved over unbalanced data, either for in-sample or for out-sample cases. The open-source software R is used for implementation of the methods.

This paper focuses attention on prediction of economic crisis using a recursive dynamic economic ... more This paper focuses attention on prediction of economic crisis using a recursive dynamic economic model. The general background of the crisis and the various financial and economic arguments to explain the crisis are included into the model. These arguments were used considering a hypothetical country with general global financial interlinks to produce a crisis prediction model. Employed in this paper are the recently discovered methods in recursive macroeconomics to build a dynamic model to explain the general financial picture of the hypothetical country. The model applied to Asian Crisis revealed that the effect of Asian Economic Crisis could last a considerable period of time and that any country could be vulnerable to economic crisis if a strategy of proper economic planning and management is not exercised. The model provides a practical tool for macroeconomists while it lays a foundation for future development of new economic models that enable one to study the state of an economy with various policy alternatives

Catalyst agent (CA) is the name given to a new concept in which artificial data points are introd... more Catalyst agent (CA) is the name given to a new concept in which artificial data points are introduced with an intension of later filtering them togenther with massive amount of corrupted data that mask possible relationship between variables. Use of CA method could sharpen and extract valuable relationships hidden in massive datasets. This paper explores the potential of CA method in the context of data mining and modeling of massive datasets is evaluated. The concept is illustrated with a simple linear regression model and CAs with influencing effects on X and Y directions. The introduction of so-called X-influencing and Y-influencing datasets allows extraction of relationships hidden in massive datasets. The potential for future improvement of the technique in multivariate context is discussed. While nothing has been reported in this area of research in the past, the method lays the foundation for a new era of future statistical research

The recursive procedure for updating the parameter estimates and computing the residuals for a si... more The recursive procedure for updating the parameter estimates and computing the residuals for a simple Linear Mixed Model is considered assuming normally distributed errors. The method is illustrated using a balanced block structure with three blocks and two treatments

International Journal of Modelling and Simulation, 2001

This article considers the general theory associated with modelling and interpretation of some no... more This article considers the general theory associated with modelling and interpretation of some nonstationary ARMA processes driven by symmetric Q stable inputs (or infinite variance signals). A set of suitable AR and MA regularity conditions is given to ensure the existence and uniqueness of regular solutions. Some explicit prediction results are also reported, as they are very useful in many practical problems that arise in science, engineering and business.

Communications in Statistics - Theory and Methods, 2015

ABSTRACT

Statistics & Probability Letters, 2008

ABSTRACT In financial modeling, the moments of the observed process, the kurtosis and the moments... more ABSTRACT In financial modeling, the moments of the observed process, the kurtosis and the moments of the conditional volatility play important roles. They are very important in model identification and in forecasting the volatility (see Thavaneswaran et al. [(2005b). Forecasting volatility. Statist. Probab. Lett. 75, 1-10]). This paper introduces random coefficient GARCH models including the class random coefficient GARCH (RC-GARCH) models and derive their higher order moments and kurtosis.(c) 2007 Elsevier B.V. All rights reserved.

Mathematical and Computer Modelling, 2002

This paper considers the modelling and analysis of nonstationary vector ARMA processes from the t... more This paper considers the modelling and analysis of nonstationary vector ARMA processes from the theoretical point of view. Some extensions to the existing work (see, for example [1,2]) incorporating the nonstationarity of the underlying process are given using models with nonstationary innovations. It is shown that this class of models provides a very strong framework for many practical situations. The estimation procedures are discussed. Various prediction results are also given. Some examples are added to illustrate the theory.

Mathematical and Computer Modelling, 2001

ABSTRACT Infinite variance processes have attracted growing interest in recent years due to its a... more ABSTRACT Infinite variance processes have attracted growing interest in recent years due to its applications in many areas of statistics. For example, ARIMA time series models with infinite variance innovations are widely used in financial modelling. However, little attention has been paid to incorporate infinite variance innovations for time series models with random coefficients introduced by Nicholls and Quinn [1]. Estimation of model parameters for some special cases are discussed using the least absolute deviation (LAD) estimating function approach when the closed form density is available. It is also shown that these new LAD estimates are superior to some of the existing ones.

Journal of Time Series Analysis, 1988

This paper considers some extended results associated with the predictors of long-memory time ser... more This paper considers some extended results associated with the predictors of long-memory time series models. These direct methods of obtaining predictors of fractionally differenced autoregressive integrated moving-average (ARIMA) processes have advantages from the theoretical point of view.