RESCON 2023

Permanent URI for this collectionhttps://ir.lib.pdn.ac.lk/handle/20.500.14444/5953

Browse

Now showing 1 - 20 of 43

Time series analysis for modeling and forecasting tea exports in Sri Lanka
(Postgraduate Institute of Science (PGIS), University of Peradeniya, Sri Lanka, 2023-11-03) Wickramasurendra, T.; Abeysundara, S.P.
Tea is the most popular and low-cost beverage in the world, next only to water and the most consumed manufactured drink in the world made from the young leaves and unopened leaf buds of the evergreen shrub Camellia sinensis. There are many tea-growing countries worldwide, and Sri Lanka is the fourth largest tea-growing country and the third largest tea exporter to the world market. The tea industry plays a significant role in the Sri Lankan economy regarding foreign exchange earnings, providing employment opportunities and being the main source of government revenue. Therefore, it is vital to predict future fluctuations in tea exports, which affect the country’s economy. This study attempted to identify appropriate ARIMA models to forecast tea exports in Sri Lanka by export quantity and export value. Twelve variables namely, Bulk Quantity, Packet Quantity, Bags Quantity, Instant Quantity, Green Tea Quantity, Total Quantity, Bulk Value, Packet Value, Bags Value, Instant Value, Green Tea Value and Total Value were accounted for the study. Initially, it was found that all the data were non-stationary by using the Kwiatkowski Phillips Schmidt Shin (KPSS) test. Therefore, the first differencing was applied, and the stationarity of the data was confirmed. The univariate time series analysis was applied for each variable, and models with the lowest Akaike Information Criterion (AIC) and Bayesian information criterion (BIC) values were used to select the best-fitted models. The Seasonal ARIMA (SARIMA) models were fitted to forecast Bulk Quantity, Packet Quantity, Bags Quantity, Green Tea Quantity and Total Quantity. The ARIMA models were fitted to forecast Instant Quantity, Bulk Value, Packet Value, Bags Value, Instant Value, Green Tea Value and Total Value. It was found that the models fitted to forecast Instant Quantity and Instant Value have Mean Absolute Percentage Error (MAPE) higher than 50%, which indicates the lower predictive ability of the models, while all the other fitted models showed MAPE below 25%, which were relatively better suited to predict the variables.
An analysis of selected variables influencing cesarean section of pregnant mothers admitted to the Ampara General Hospital
(Postgraduate Institute of Science (PGIS), University of Peradeniya, Sri Lanka, 2023-11-03) De Zoysa, G.I.C.L.; Haalisha, M.A.
Improving cesarean section as a procedure of safety with notably low fetal and maternal mortality rates is one of the most crucial developments in modern perinatal. Most of the local population in the Ampara region predominantly rely on agriculture and fishing; therefore, there may be some gaps in understanding of child safety. Hence, studying the most influencing factors for caesareans is vital to both medical professionals and the local population. Here, the study reported the most influential factors causing Caesareans in the Ampara region using a Binary logistic regression model from secondary data of 224 observations. The results indicated that the emergency cesarean birth patients were in the 20-29 age group. While fatal distress, lack of progress, and dribbling variables had a positive influence, mother’s age negatively influenced emergency cesarean section. The patients aged 20-29 were 1.0902 more likely to develop emergency cesarean section than other age groups (Odds ratio = 0.9173); patients with fatal distress were more likely to develop emergency cesarean birth than the patients without fatal distress (Odds ratio = 26.9746), patients with a lack of progress had more risk to develop emergency cesarean birth than the patients without lack of progress (Odds ratio= 32.5027), and the patients with dribbling of emergency cesarean had the more risk to develop emergency cesarean birth (Odds ratio = 13.2761). Delivery by cesarean birth is a difficult health issue and costly. Efforts should be initiated at Ampara General Hospital to reduce cesarean sections. This study concluded that the mother’s age, fetal distress, lack of progress, and dribbling were significant predictors of cesarean birth and, therefore, must be considered when addressing the reduction of cesarean rates in Ampara General Hospital.
Convergence of the finite difference method for an age structured two-sex population dynamics model of thalassemia transmission
(Postgraduate Institute of Science (PGIS), University of Peradeniya, Sri Lanka, 2023-11-03) Chandrasiri, A.M.P; De Silva, T.H.K.R.
Thalassemia is an inherited blood disorder that affects the production of haemoglobin and red blood cells. Symptoms include anaemia, jaundice, chest pain, and breathing problems. It is an autosomal recessive disorder, meaning both parents must have the disease or be carriers for it to be passed down to the next generation. Thalassemia is native to a wide but restricted geographical area. Nevertheless, migration is spreading to formerly unaffected areas. Consequently, tracking and forecasting disease prevalence is important for effective healthcare planning. Structured models are essential for studying multicellular organism populations and hereditary diseases in which age and sex play a role. The previously published work established and analysed a two-sex age-structured continuous type population dynamics model for thalassemia transmission that describes the genotype composition of the population resulting from the thalassemia trait and is based on the Fredrickson-Hoppensteadt model, which is a system of semi-linear partial differential equations with nonlocal boundary conditions. To make projections about a population, we must generate numbers from them to compare with data. The model’s answer must then be approximated using numerical methods. The objective of this research is to present a numerical algorithm for approximating the solution of the model and to demonstrate that this method converges ideally to the exact answer. The Crank-Nicolson form of the finite difference method of characteristics, combined with the trapezoidal rule for the quadrature of the integrals that describe births and densities of married individuals of each sex, is developed to approximate the solutions of the model. The optimal rate of convergence of the numerical method is discussed to the maximum norm. The work presented here can potentially be beneficial in both mathematical and biological contexts.
Combining fast text embeddings with neural networks for short text classification
(Postgraduate Institute of Science (PGIS), University of Peradeniya, Sri Lanka, 2023-11-03) Jayakody, J.R.K.C.; Vidanagama, V.G.T.N.
Using embedding representation is a critical step to improve the classification accuracy of a text dataset. Even though Bag of Word (BOW) models are used with past research work, usage of word2vec, Glove and FastText as embedding techniques helps to represent the features of text documents in a distributed manner, hence improving the accuracy of such models. The latest research work used a combination of embedding techniques and enhanced neural network models to improve the classification accuracy of text documents. FastText as an embedding unsupervised model and CNN, LSTM, and RNN as neural models were used extensively in the latest research work. However, comprehensive analysis with FastText and neural models with text documents has not been undertaken thus far. As a result, it is hard to compare the existing research work, and it is unclear which combination of neural model with FastText performs well over the other techniques. Therefore, it is necessary to investigate the impact of neural networks when the features were represented with the FastText embedding model. A famous movie review dataset was used for the experiment. CNN, LSTM, RNN, NN, and variations of those neural networks were used as neural networks. Hold out stratified Training and testing set was taken with 70 % to 30% split. Seventy per cent of training data was split as 80% of training and 20% of validation set. We compare classification accuracy across a range of neural network models, and our results show that the RNN model outperforms other neural network models with FastText embeddings with 86% accuracy. Moreover, out of various neural networks, the combination CNN-LSTM outperforms all other neural network models with 88% accuracy. The outcomes of this study can be a baseline for future research.
Generalised lambda distribution-based quantile regression model to analyse the exchange rate movements in Sri Lanka
(Postgraduate Institute of Science (PGIS), University of Peradeniya, Sri Lanka, 2023-11-03) Basnayake,B.R.P.M.; Chandrasekara, N.V.
Exchange rates play an important role in currency trading in the financial market. Hence, this study examines the relationship and forecasts close price values of USD, EURO and GBP against LKR using the previous day’s low, high and open price values, lags of close price and moving average values. This is the first study that implements quantile regression (QR) incorporating Generalised Lambda distribution (GLD) to model exchange rates due to the non-normal behaviour of the residuals with the presence of heteroscedasticity in Sri Lanka. Daily data was collected from the Yahoo Finance website from 1ˢᵗ January 2008 to 28ᵗʰ February 2022. The current daily close price of exchange rates was modelled using the previous day’s low, high and open price values, lags of close price observed in the Autocorrelation function (ACF) plot and moving average (MA) values of MA7, MA14, MA28, MA84, MA168 representing the moving average values for one week, two weeks, one month, one quarter and six months respectively. For heteroscedastic data, QR models were obtained by Case I) fixing the intercept or Case II) allowing all the coefficients to vary using the Nelder-Mead simplex algorithm. The Cramer-Von Mises and Anderson-Darling tests were used to evaluate whether the residuals follow a GLD in the GLD-based QR models. Further, the goodness of fit of these QR models was evaluated using Pseudo-R² . This study considered upper, median, and lower quantiles in fitting the QR models. The forecasted accuracy of the QR models was evaluated using mean absolute error (MAE) and mean absolute percentage error (MAPE). It is found that the effects of the input variables on the close price of the exchange rate at different quantiles are different. The minimum MAE and MAPE of 1.3246 and 0.00587 were observed for the 50th QR model with a Pseudo-R² value of 0.4432 in EURO/LKR, for USD/LKR minimum error values (MAE of 1.1369 and MAPE of 0.0057) were observed under Case I in 50th QR model with Pseudo-R² of 0.7541. Similarly, for GBP/LKR, the better-performed model was under Case I in the 10th QR model (MAE of 1.2253 and MAPE of 0.0045) with Pseudo-R² of 0.9255. Overall, this study indicates that QR models can emphasise the complete conditional distribution of the response variable at different quantiles.
A new method using the geometric mean to solve tetrad fuzzy transportation problems
(Postgraduate Institute of Science (PGIS), University of Peradeniya, Sri Lanka, 2023-11-03) Ekanayake, E.M.T.D.K; Ekanayake, E.M.U.S.B.
Globally, the transportation problem is employed in the concrete obstacles. A transportation problem is crucial for the production business, where optimisation techniques are functional for solving multi-objective problems and network flow analysis. However, in real-world problems, Fuzzy Transportation problems (FTP) are accurate and widely used in engineering applications and fields such as Operation Research, management science, and control theory. The main goal of this research is to determine the lowest transportation cost of moving certain goods through a capacitated network where supply and demand for nodes, as well as the capacity of edges, are represented as tetrad (trapezoidal) fuzzy numbers. The ranking method is one of the most common methods for solving fuzzy transportation problems (FTP). Instead of using the ranking method to get the best solution to the FTP, a new method is proposed using geometric means to solve a tetrad FTP, where demand and supply are all represented as tetrad fuzzy integers. This approach is easy to understand and applicable to real-life transportation problems.
The role of Bilingualism in the revival of a dead language: manx in isle of man
(Postgraduate Institute of Science (PGIS), University of Peradeniya, Sri Lanka, 2023-11-03) Kariyapperuma, C.P.; De Silva, K.P.
The survival of many languages in the modern world is at risk due to the quick pace of globalisation and the dominance of a few languages. To explain language dynamics in the context of the aforementioned problem, the academic community has used mathematical models. In particular, the literature has attempted to explain the disappearance of languages like Manx, Celtic, Gaelic, Welsh, and Quechua by focusing on how more advantageous languages draw people and decrease the number of speakers of less advantageous languages. This study focuses on the decline and subsequent revival of the Manx language on the Isle of Man, which was introduced by Irish raiders around AD 500. Historical accounts suggest that the shift from Manx to bilingualism occurred by the early 19th century, and the transition from Manx to English primarily took place throughout the 19th century, with a significant acceleration towards the end of the 19th and beginning of the 20th century. The death of the last native Manx speaker was reported in 1974, causing the extinction of the informal transmission ways of the Manx language. However, efforts were made to revive the language in schools and educational institutions, leading to Manx being categorised as a revitalised language. To understand this scenario, the study employs a three-dimensional language model proposed in a previous work that incorporates the concept of bilingualism in a wellmixed bilingual society. The model was tested on the real-world data of the population fractions speaking Manx from 1901 to 1974. The results suggested that a ‘hidden’ bilingual minority (speaking both Manx and English) must have persisted despite the declared extinction of Manx in 1974, possibly due to the stigma attached to speaking the language. This bilingual minority may have contributed to the recent revival of Manx.
Manufacturers and buyers integrated production inventory model with better synchronisation
(Postgraduate Institute of Science (PGIS), University of Peradeniya, Sri Lanka, 2023-11-03) Hisam, M.S.M.; Daundasekera, W.B.; Rodrigo, W.N.P.
Coordinating inventory between multiple manufacturers and buyers is a common practice in the current economic environment. It gives several advantages for both manufacturers and buyers. In the literature, more studies focused on single manufacturer and single/multi buyer(s) integrated production inventory models with various factors. However, limited studies have investigated multi-manufacturer and multi-buyer scenarios. This study investigates multi-manufacturer and multi-buyer integrated production inventory systems. Here, we assume manufacturers produce a homogeneous product and deliver a lot just after its production. Moreover, manufacturers supply their products to all buyers to satisfy buyers’ demands. A multi-manufacturer and multi-buyer integrated production inventory model was developed by considering realistic factors and a combination of equal and/or unequal size batch transferring policies. It is assumed that batch sizes follow geometric series whose common ratio is less than 1—subsequently, an optimal solution technique to the proposed model was derived to obtain minimal total cost. Finally, a real-world problem is used to illustrate the analytical findings of the study.
Explicit finite difference method for the valuation of American put option with dividends using logarithmic front fixing transformation
(Postgraduate Institute of Science (PGIS), University of Peradeniya, Sri Lanka, 2023-11-03) Selvarajah, P.; Kajanthan, S.
The most traded options are American options due to their early exercise features. An American option is a type of financial derivative contract that gives the holder (buyer) the right, but not the obligation, to buy (call option) or sell (put option) an underlying asset at a pre-specified time before the expiration date of the option. However, pricing American options has an intriguing problem. The early exercise behaviour makes the American option pricing problem become a free boundary problem. Logarithmic front fixing transformation is a class of transformation used to transform the American free boundary problem into a fixed boundary problem. Numerical methods are used to solve the transformed fixed boundary problem. The effects of dividend payments on American put option pricing valuation were studied, and the explicit finite difference method was used to obtain the numerical solution. Dividends can significantly impact the price of an underlying asset and, consequently, the pricing of American options. When a company pays a dividend, it reduces the value of the stock because the cash is transferred to the shareholders. For put options, which give the holder the right to sell the underlying asset, dividends can increase the option’s value because the holder is protected from the drop in stock price caused by the dividend payment. The inclusion of dividends guarantees stability by the numerical experiments, and numerical results confirm that increase in the American put option price and the optimal exercise price. Thus, dividend yields early exercise of the American put option less likely.
Fuzzy inference system to identify disaster risk levels in Sri Lanka
(Postgraduate Institute of Science (PGIS), University of Peradeniya, Sri Lanka, 2023-11-03) Karunarathne, A.W.S.P.; De Silva, K.; Perera, S.S.N.
Various uncontrollable factors influence natural disasters. Natural events, for instance, floods, occur during periods of extreme rainfall and can pose significant challenges and risks to Sri Lanka’s agricultural sector. They threaten crop production and overall food security in the country. It is crucial to identify potential flood and drought threats related to rainfall to mitigate these risks. The Standardised Precipitation Index (SPI) is used to measure rainfall, but it alone cannot determine the risk levels for floods and droughts because it lacks clear definitions for prolonged periods and disaster thresholds. In this study, we have developed a fuzzy expert system that assesses the impact of rainfall fluctuations, as measured by the SPI, on the Disaster Risk Level (DRL) while considering uncertainty through fuzzy membership functions. This decision support system is a single-input, single-output model, with the SPI as the input variable and DRL as the output variable. Linguistic terms like Extremely Dry, Dry, Moderate, Wet, and Extremely Wet are used to describe SPI categories, while No Disaster, Disaster, and High Disaster are used for DRL. We calibrated this fuzzy expert rule-based system using historical records of floods and droughts in 1983, 2003, 2011, 2016, 2017, and 2020. The resulting fuzzy inference-based decision support system, which evaluates DRL based on SPI data, offers a practical and valuable solution for proactive disaster management and preparedness.
Comparison of the predictive performance of liu-based estimators
(Postgraduate Institute of Science (PGIS), University of Peradeniya, Sri Lanka, 2023-11-03) Kayathiri, T.; Kayanan, M.; Wijekoon, P.
Logistic regression is one of the most widely used statistical methods for predicting a binary outcome by analysing the relationship between one or more existing independent variables. Although the maximum likelihood estimation technique is a commonly used method to estimate the parameters, their predictive performances may be affected by a problem called multicollinearity. To reduce the effect of multicollinearity, different biased estimators have been proposed as alternatives to the Maximum Likelihood Estimator (MLE). According to the literature, the superiority of the existing estimators based on Liu estimators was examined using the Mean Square Error Matrix (MSEM) and Scalar Mean Square Error (SMSE) criteria. However, the researchers did not compare the prediction performance of these existing estimators. Therefore, the present study is aimed to compare the prediction performance of Liu-based estimators in logistic regression using balanced accuracy. The prediction performance of the Maximum Likelihood Estimator (MLE), Logistic Liu Estimator (LLE), Almost Unbiased Liu Logistic Estimator (AULLE), and Modified Almost Unbiased Logistic Liu Estimator (MAULLE) are considered for comparison. To evaluate the balanced accuracy of the above estimators, the dataset was split into two so that 70% belongs to the training set and 30% to the test set. The model was trained using the training set, then the testing set was used to evaluate the balanced accuracy. A Monte Carlo simulation study was done to understand the prediction accuracy by setting different levels of correlation among the predictors and sample sizes. Further, a myopia real-world dataset was utilised, and it was observed that the related results tally with the results of the simulation study. Finally, it was noticed that the estimator MAULLE has the best prediction performance when multicollinearity is present, and then LLE performs well. Additionally, the prediction performance of AULLE was significantly better for some selected values of shrinkage parameters. However, the MLE performs well with small sample sizes.
Decision tree algorithms to determine GCE Ordinary-Level student performance factors influencing English as a subject
(Postgraduate Institute of Science (PGIS), University of Peradeniya, Sri Lanka, 2023-11-03) Pinnawala, H.A.K.B.
The General Certificate of Education Ordinary Level (GCE O/L) Examination is a milestone in the Sri Lankan Education system. The GCE O/L results are used to screen students for selecting subjects for the Advanced Level Examination. Students’ learning strength depends on many factors, such as environmental, mental, and physical factors. Since these factors directly affect the student’s learning ability, it is crucial for educators to identify the most important factors among these. Randomly selected student data were used in this study. The questionnaire consisted of questions that were assumed to affect the outcome of the English subject result. Seventy-two Grade 12 students answered the questionnaire. Various decision tree algorithms were used for the classification. J48, LMT and Random Tree were used to build the classification models using WEKA, and their accuracies were compared. During the study, English writing reading ability, family help and contribution of tuition classes showed an association with the student grades. After the model creation, J48, LMT and Random Tree obtained 51%, 47% and 55% accuracy, respectively. The decision tree model with the highest accuracy was then considered, and the decision tree classification rules were converted manually into If-Else statements. Random Tree obtained the highest accuracy. Using the classification rules generated by the Random Tree model, the performance factors related to English as a subject were identified.
On ℎ-function of a bounded simply connected region: disc with deleted double slits
(Postgraduate Institute of Science (PGIS), University of Peradeniya, Sri Lanka, 2023-11-03) Arunmaran, M.
The h-function or harmonic-measure distribution function for a region 𝛺 with a specific fixed point zₒ 𝛺, gives some information about the shape of the region. For a region 𝛺 with a basepoint zₒ, we identify the set 𝐸ᵣ, which is the intersection of the boundary of 𝛺 and the closed disc of radius 𝑟 centred at zₒ. The ℎ-function ℎ(𝑟) is given by the harmonic measure of the set 𝐸ᵣ, in 𝛺 at zₒ. This function ℎ(𝑟) only takes the values in the unit interval [0,1], but ℎ(𝑟) will take the value one only for the regions with bounded boundaries. This study is focused on the behaviour of the ℎ-function(s) of a bounded simply connected regions 𝛺 formed by deleting double slits from the unit disc centred at the origin when both slits lie on the real axis and vary in length. Three cases are considered: keep the size of both slits the same; keep the length of one of two slits as the radius of the disc; keep the length of one of two slits as it is bigger than the radius of the disc. For these regions 𝛺, the ℎ-function will take the value 1 after some values of 𝑟. That is, the ℎ-function ℎ(𝑟) meets the line 𝑦= 1 at an angle 𝜓, subtended at the line 𝑦 = 1 with the graph of ℎ(𝑟) in the counterclockwise direction. We check how this angle 𝜓 changes for the above three cases when the basepoint zₒ lies anywhere in between both slits inside the disc. For the first and third cases, when the basepoint zₒ moves between two slits from left to right along the real axis, the angle 𝜓 increases from zero for a while and attains its maximum and then decreases to zero. In the first case, the maximum of the angle 𝜓 has been attained at 𝜋/2. For the second case, when the basepoint zₒ moves between the two slits from left to right along the real line, the angle 𝜓𝜓 decreases from 𝜋/2 to zero. These findings indicate that the ℎ-function of these bounded regions 𝛺 has interesting behaviour at the point 𝑟∗ which is the furthest distance between the base point and the boundary of the region. Future research will focus on checking the behaviour at the same point 𝑟∗, when the basepoint varies along the imaginary axis within these regions 𝛺.
Ensemble learning approach for youtube video classification based on their video content
(Postgraduate Institute of Science (PGIS), University of Peradeniya, Sri Lanka, 2023-11-03) De Siva, N.H.T.M.; Rupasingha, R.A.H.M.
Many individuals use the platform YouTube to share videos with others worldwide. Different content and quality videos are available on the YouTube platform, and it would be beneficial to users if they could assess the content of the video before selecting a video to view. However, deciding and making predictions directly is difficult since different attributes need to be considered. This study is carried out to classify videos using six individual distinct machine learning algorithms and an ensemble learning algorithm as a solution to the above. The algorithms were selected based on the literature review. The gathered data set underwent essential pre-processing and attribute ranking. View count, like count, comment count, number of subscribers, tag count, and total views were identified as the main contributors to the study. For classification, Naive Bayes, Logistic Regression, Support Vector Machine (SVM), Decision Tree, Multilayer Perception (MLP), Random Forest individual algorithms, and Ensemble Learning algorithm that combined five individual algorithms were selected since those algorithms work effectively and efficiently with better results. Among the ensemble learning algorithm techniques, majority voting exhibits the best accuracy. For categorisation, the result was evaluated by 60% training data and 40% testing data. In each method, different parameters were changed for the evaluation and accuracy, recall, f-measure, precision, Root Mean Square Error, and Mean Absolute Error were taken into account. Random Forest demonstrated 96.89% accuracy, and Ensemble Learning showed 97.21% better accuracy than others. These findings indicate that this strategy is appropriate for YouTube content classification. It offers the information required to evaluate video content and make predictions about particular videos. Future research would focus on applying more deep learning algorithms to improve accuracy.
A memetic algorithm for the vehicle routing problem with moving shipments at the cross-docking warehouse
(Postgraduate Institute of Science(PGIS), University of Peradeniya, Sri Lanka, 2011-11-03) Gnanapragasam, S.R.; Daundasekera, W.B.
Cross-docking (CD), a relatively new technique, is considered an efficient technique to control the inventory flow in logistics. CD can reduce delivery lead times, inventory holding, and transportation costs. This research extends the study on vehicle routing problem with moving shipments at the cross-docking center (VRPCD&MS). The objectives of this study are to obtain near-optimal solutions to the large-scale instances of the VRPCD&MS model using a meta-heuristic algorithm known as the Memetic Algorithm (MA) and compare the performance of MA with the Genetic Algorithm (GA). In this study, MA is hybridised with GA, with an insertion local search method. The elitism method to choose the best members from the previous population to the next is also considered to structure the proposed MA approach. The tournament selection, order crossover and swap mutation are applied as the operators of the GA. The data for the numerical experiments are extracted from a benchmark problem in the literature. At the preliminary analysis, some parameters of MA, such as population size, number of iterations, termination count, crossover rate, and mutation rate, are tuned by the Taguchi method, and the appropriate parameter values are 50, 100, 100, 0.7, and 0.3, respectively. The computational results show that better solutions are found for VRPCD&MS by the MA approach than GA. In all the instances, even the average solutions found by MA are better than the best solutions found by the GA approach. Also, it was observed from the convergence analysis that the MA approach can reach the solution in fewer iterations than the GA approach. Therefore, it can be concluded that MA is capable of providing more accurate solutions than GA, whose average percentage improvement is nearly 6%. Moreover, it can be concluded that the MA approach converges to a better nearoptimal solution faster than the GA approach.
Assessing agreement between two measurement systems using replicated scale mixtures of skew-normal measurement error models with varying degrees of freedom
(Postgraduate Institute of Science (PGIS), University of Peradeniya, Sri Lanka, 2023-11-03) Duwarahan, J.; Nawarathna, L.S.
Method comparison studies are commonly conducted in health fields to evaluate the interchangeability of a new method for measuring a continuous variable with an established reference method. The agreement between two methods that measure the same variable but are prone to measurement errors is often evaluated by measurement error models, which are assumed to be normal. However, normality may not hold when dealing with skewed and heavy-tailed data. To address this issue, a replicated measurement error model (RMEM) is proposed for analysing replicated method comparison data with different levels of heaviness in the tails of true covariates and errors under scale mixtures of skew-normal (SMSN) distributions. The model, which includes skew-𝑡𝑡 (ST), skew generalized-𝑡 (SGT), and skewslash (SS) distributions, is called generalised scale mixtures of skew-normal RMEM (GSMSN-RMEM). The proposed methodology is evaluated through a simulation study using root mean square error measures for sample sizes of 𝑛 = 50, 100, and 200, and the expectation conditional maximisation approach is applied to fit the model. The simulation results indicate that ST and SGT distributions outperform the skew-normal distribution, possibly due to their heavy-tailed characteristics. Furthermore, the methodology is demonstrated by analysing systolic blood pressure data, and model selection is employed using the Akaike information criterion and Bayesian information criterion. The agreement between methods is assessed using the unconditional probability of agreement, and it is found to be higher for SGT (nearly 0.95) and ST (nearly 0.90) distributions compared to other distributions. The study demonstrates that the proposed method, GSMSN-RMEM under ST and SGT distributions, is an effective tool for evaluating the agreement between two measurement systems when dealing with measurement errors and skewed and heavytailed data. This method can be applied in various fields, such as biomedical engineering, clinical research, and medical imaging.
A comparison of bird diversity between the core area and the buffer zone of Bodhinagala Forest Reserve, southwestern Sri Lanka
(Postgraduate Institute of Science (PGIS), University of Peradeniya, Sri Lanka, 2023-11-03) Madusanka, W. N.
Bodhinagala Forest Reserve (BRF) is located in Dhombagaskanda, Ingiriya, within the Kalutara District of south-western Sri Lanka. Based on the floristic composition and structure, it can be defined as a lowland rainforest surrounded by well-wooded home gardens of Bhodinagala village. A rich avifauna has been documented by previous studies conducted in BRF. A thirty-month-long avifaunal survey was conducted from October 2020 to April 2023 using a visual encounter survey with bird calls, and point counts were conducted along a 2,500 m long line transect with 150 m gaps between consecutive points. Birds surveys were conducted between 0600 h and 1000 h, 1400 h and 1800 h. Bird diversity was measured using Shannon (H′) and Simpson’s (1-D) diversity indices. A total of 314 birds consisting of 82 species belonging to 19 orders and 46 families were recorded in both transects. The Shannon Index and Simpson’s Index, respectively, for the core forest area were 3.91 and 0.97, and for the buffer zone, were 3.96 and 0.97. Thirteen endemic bird species were recorded, such as Sri Lanka Swallow (Cecropis hyperythra), Red-backed Flameback (Dinopium psarodes), Sri Lanka Junglefowl (Gallus lafayettii), Green-billed Coucal (Centropus chlororhynchos) and Sri Lanka Gray Hornbill (Ocyceros gingalensis), while migrant birds such as Barn Swallow (Hirundo rustica), Blue-tailed Bee-eater (Merops philippinus), and Asian Brown Flycatcher (Muscicapa dauurica) were also documented during the study. Species such as Malabar trogon (Harpactes fasciatus), Black-backed Dwarf-Kingfisher (Ceyx erithaca) and Black-naped Monarch (Hypothymis azurea) are uncommon in the lowland wet zone, were frequently observed. Both areas of the forest reserve and buffer zone showed equally high bird diversity. This study identified a rich avifaunal diversity that has not been extensively studied in this reserve. We recommend conducting comprehensive biodiversity surveys to identify potential threats and develop conservation strategies for both the species and their habitats.
Fish assemblage structure and associated environmental factors of Rawan-Oya tributary of Mahaweli river
(Postgraduate Institute of Science (PGIS), University of Peradeniya, Sri Lanka, 2023-11-03) Weerakoon, K.C.; Abayasekara, C.L.; Kapukotuwa, G.K.; Rajakaruna, R.S.
Local assemblages of organisms in aquatic ecosystems are associated with environmental factors that determine water quality. This study assessed the fish assemblage structure along the Rawan-Oya Tributary of the Mahaweli River and its association with physicochemical properties and faecal indicator bacteria (FIB) levels. Study sites represented forested, agricultural, rural, semi-urban and urban areas along the river. Within the sites, habitat types were defined as run, riffle, and pool. Fish species were sampled from February 2020 to May 2022, covering wet and dry seasons and identified in situ. Physicochemical parameters and FIB were assessed with standard protocols. Fish Species Richness (S) and the Shannon Diversity Index (H’) were calculated using R statistical software. A canonical correlation analysis identified the relationship between fish assemblage structure and water quality parameters. Twenty fish species belonging to 17 genera and 11 families were recorded. The family Cyprinidae was the most dominant, followed by Poeciliidae and Danionidae. Fish species Dawkinsia singhala, Garra ceylonensis and Schistura notostigma are endemic to Sri Lanka. The characteristics of the habitat strongly influenced the fish assemblage structure. The species richness (S) and diversity (H’) in pool, run, and riffle habitats were reported as 19, 19, and 5 and 2.5, 2.3 and 1.5, respectively. There was no difference in fish diversity and richness between dry and wet seasons (GLM; p > 0.05). Schistura notostigma was the most influential species at high altitudes (canonical coefficient = 3.7) and was associated with high dissolved-oxygen content, low levels of nutrients and FIB levels. Poecilia reticulata, and Devario malabaricus were common in sites with high biological oxygen demand, electrical conductivity, high faecal and nutrient pollution and demonstrated high tolerability towards the reduced water quality conditions. The fish assemblage structure of the tributary exhibited relationships with the habitat characteristics and the water quality parameters.
A web-based landslide risk dissemination portal incorporating bayesian probabilistic risk prediction mechanism on landslide causative parameters
(Postgraduate Institute of Science (PGIS), University of Peradeniya, Sri Lanka, 2023-11-03) Gammanpila, G.H.D.T.N.; Rodrigo, U.H.G.; Weerakoon, I.T.; Welikanna, D.R.
Landslides severely threaten the environment, human life, and infrastructure in hilly regions worldwide. Accurate prediction and identification of landslides are crucial for effective risk management. This study utilises Bayesian probabilities and machine learning techniques with geospatial data analysis to develop a reliable model for landslide identification in Ratnapura district in Sri Lanka, known for its high landslide risk. The study utilised data sources, including SAR images, rainfall data, slope data, aspect data, and land use data, and processed the collected data. Processed parametric data were integrated into a Bayesian probabilistic model. The landslide risk map was created using these probabilistic values to classify the study area into different risk levels. The validation of the Bayesian probabilistic model using data from the NASA Landslide Inventory Catalogue confirms its accurate prediction of risk levels for landslide-occurred locations and known low-risk areas, demonstrating its effectiveness in assessing landslide risk. A machine learning model has been successfully implemented to establish a relationship between rainfall data and geospatial landslide risk, producing an output that accurately reflects their connection. The model demonstrated exceptional performance, achieving a training set accuracy of over 98% and a perfect 100% accuracy on the test set. The developed model was integrated into a userfriendly web application that government agencies and the general public can use to identify high-risk landslide areas. This tool can potentially improve landslide risk management practices in hilly regions worldwide by providing valuable information to stakeholders and decision-makers so they can make informed decisions regarding risk management and emergency response measures. The findings of this research were disseminated through a real-time GIS web application, which facilitated the dissemination of information regarding high-risk areas for landslides to minimise the devastating impact of landslides on communities and infrastructure.
Does shelf life affect the level of microplastics in bottled water
(Postgraduate Institute of Science, University of Peradeniya, Sri Lanka, 2023-11-03) Premathilake, B.A.P.C.; Kapukotuwa, G.K.; Rajakaruna, R.S.
Microplastics (MPs) have been detected worldwide in freshwater systems and drinking water. This longitudinal study examined the effect of the shelf-life of bottled water on MP levels and their properties over time. Bottled water samples were purchased from the market in the Kandy District and were stored at room temperature with ambient light conditions for 3, 6, and 12 months. Water was then filtered through a 0.45μm membrane filter to separate MPs. The membrane filter was examined and enumerated under the stereomicroscope at 40X magnification. The MPs were classified according to their colour, form, and shape. Confirmation of detected particles as MPs and identification of polymer type was performed using FTIR spectroscopy. Of the 44 bottles analysed, 43 (98%) had MPs, which were mostly found as fibres, followed by films and fragments. About half of the MPs were transparent; others were blue, pink, black, brown, purple and yellow. Twelve types of polymers and sizes ranging from 5 – 5,000μm were identified. There was no difference in the concentration of MPs in bottled water with different storage durations (Two-way ANOVA; f = 14.54; p > 0.05). However, the number of MPs significantly decreased with the storage time, from a range of 2–28, 2–20 and 5–12 MPs/L in bottles kept for 3, 6, and 12 months, respectively. The results showed that although the shelf life did not increase the amount of microplastics in bottled water, they may break into smaller nanoplastics during longer storage, which has to be further investigated.

Browse

Recent Submissions