Research & Projects

Post-doctoral level research on Correlated Cluster Data

Prof.(Mrs) M.R. Sooriyarachchi

Correlated clustered data occur in many fields particularly in Medicine, Biology and Social Sciences. Due to the cluster correlation, simple methods of analysis cannot be used as the correlation has to be adjusted for. There are three alternative methods of analyzing this type of data, namely, Multilevel Modelling, Generalized Estimating Equation Methods and Modified sandwich variance methods. Developing methods for multivariate responses at different cluster levels are a matter of interest. Methods will be applied to Sri Lankan Dengue data. The output from this research will be two fold. There will be methodological developments in the field of Statistics (Pure Research – Natural Science) and applied research in the field of Health Science. An undergraduate working in this area with me has produced a paper accepted by an international journal.

A Mortality Forecasting Model for Sri Lanka

Dr. W.N. Wickremasinghe and Prof. Lakshman Alles (School of Economics and Finance, Curtin University, Australia)

Sri Lanka like many other countries in the world, currently has an ageing population and increasing life expectancy. The global financial crisis also had an impact on the financial security of retirees. A major decision retirees have to make at the time of retirement is the structuring of the wealth pool accumulated up to that point in the best possible manner to provide for their life during retirement. However, there are some factors uncertain by nature, to be considered in designing of savings and income products for the benefit of retirees. If the remaining life span is known, a financial plan can be drawn to convert the asset pool to an income stream over that period. Unfortunately, the life expectancy of a retiree is a very uncertain factor by nature. It depends on the gender and it can also change with time. It is therefore important to estimate it accurately for females and males separately and for specific age groups under normal circumstances, using appropriate statistical models.

A final year research project in Industrial Statistics was completed in August 2011 by Wasana Aberathna under the joint supervision of the above researchers, which sought to develop a statistical model for estimating and forecasting mortality rates and life expectancy rates for Sri Lanka. This project is now due for submission. A major task of the project was to hand collect the available age-specific death rates for females and, males, for 1950 – 2008 for which the data was available from the Registrar General’s office, and fill data gaps using statistical techniques. Then the objective was to use modern statistical modeling techniques to estimate death and mortality rates, and to build up accurate life tables for Sri Lanka. The project has been successful to a large extent in the face of many practical difficulties. This project will be extended to a wider project that seeks to investigate and develop a model for forecasting the financial needs of retirees in Sri Lanka.

A Study of S & T Doctoral Degree Holders in Sri Lanka and Their Contributory Patterns

Dr. W.N. Wickremasinghe and Dr. (Mrs) Seetha I. Wickremasinghe (NSF)

With the global trend in moving towards a knowledge- based society, the educational qualifications and trained skills of personal become important since there is a strong link among postgraduate education, research and development (R & D), and economic growth of a particular country. The government policy document “Mahinda Chinthanaya” highlights several areas that are crucial for the fast development of the country where R & D in the Science and Technology sector play a major role.
A Masters Project in Applied Statistics was successfully carried out by W.A.L.D.C.Weerasooriya of the NSF, Sri Lanka under the guidance of the above researchers with the objective of identifying contributory patterns of S & T doctoral degree holders toward production, innovation and transmission of R & D knowledge where policy intervention assumes important. The project was based on data from the 2008/2009 tracer study of S&T doctoral degree holders in Sri Lanka, conducted by the S&T Policy Division of the National Science Foundation. A statistical approach using descriptive techniques and categorical methods, was adopted in the project.
( Thesis was successfully defended at the viva in August 2011 and it’s pending award of degree).

Identification of Suitable Paddy Varieties for Diverse Agro-Ecological Environments in Sri Lanka

Dr. W.N. Wickremasinghe and Mr. R. A. B. Abeygunawardana

In order to increase the rice production in a small but highly populated country like Sri Lanka, it is necessary to increase productivity and the cropping intensity. This is usually achieved by introducing new and high yielding varieties cultivated in both Yala and Maha seasons.
W. M. P. M. Wickramasinghe of the Wayamba University successfully completed a Masters Project in Applied Statistics in 2009 under the guidance of the above researchers with the objective of identifying suitable and adaptable rice varieties under diverse agro-climatic conditions. The project was based on data from a genotype-environment trial of rice, conducted by the National Coordinated Rice Variety Testing (NCRVT) program of the Rice Research Institute, Batalagoda. A statistical approach relevant for analyzing genotype-environment interaction studies, was adopted in the project.

[Publications:
W.M.P.M.Wickramasinghe, W.N.Wickremasinghe, and R.A.B.Abeygunawardana (2011).
“A detailed analysis of a genotype-environment study on rice in Sri Lanka”, in Proceedings of the Annual Research Symposium of University of Colombo (extended abstract only).]

Customer Satisfaction Survey – Sri Lankan Airlines (2010)

Dr. M.D.T. Attygalle
The project requires investigating the current status of SriLankan Airlines service standards and identifies services that can be enhanced to attract new customers and improve customer retention through improved customer satisfaction. The team requires carrying out appropriate studies to capture all services and service points of SriLankan Airlines customers. A comprehensive report based on the SriLankan customer services with recommendations must be submitted as the final product of the survey.

A feasibility study for Designing & Implementing a Pilot Citizen Report Card (2011)

Dr. M.D.T. Attygalle
A Core Group member in the Designing & Implementing a Pilot Citizen Report Card for the Local Economic Growth Project jointly carried out by The Asia Foundation, SPARC, University of Colombo and the Department of Statistics, University of Colombo. Through such studies, the Users of any public service can make an impact on the quality of services offered to them.

Research Collaboration

Dr. C. D. Tilakaratne

As a result of ongoing research collaboration with Dr. M. A. Mammadov (Senior Research Fellow, School of Science, Information Technology and Engineering, University of Ballarat, Australia, two conference papers, one journal paper and one book chapter were published during the last four years.
Planning start another collaborative research project (title: An Improved Probabilistic Neural Network Algorithm for Directional Prediction of a Stock Market Index) with Dr. Mammadov. A student who is going to register at the University of Colombo will carry out this research study under the supervision of Dr. Mammadov and myself, as her PhD research.