PGIS Research Congress

Permanent URI for this community

https://ir.lib.pdn.ac.lk/handle/20.500.14444/5

Browse

Browsing PGIS Research Congress by Author "Abeysundara, S. P."

Now showing 1 - 5 of 5
  • Thumbnail Image
    Item
    Bird species classification using statistical feature analysis of avian flight calls
    (Postgraduate Institute of Science (PGIS), University of Peradeniya, Sri Lanka, 2024-11-01) Mahanayake, K. M. L. K.; Abeysundara, S. P.
    Detecting bird species through avian flight calls is essential for monitoring biodiversity and assessing the sustainability of ecosystems. This method provides a non-invasive way to study bird populations and their migratory patterns, aiding conservation efforts and ecological research. This study aimed to explore three research questions: the effectiveness of statistical feature extraction methods in bird identification, the classification of birds using supervised machine learning methods, and selecting the best model that classifies the birds. The study used the CLO-43SD dataset with multi-class species identification in avian flight calls. The dataset consists of 5428 audio clips of avian flight calls from 43 different species of North American wood warblers in the family Parulidae and is split into training and testing sets for analysis and validation. The analysis is conducted using R and Python. Statistical features such as the number of samples, length, root mean square error (RMSE), zero crossing rate, chromatogram, tempo, spectral centroid, spectral contrast, and Mel frequency cepstral coefficient (MFCC) were extracted for each avian flight call. These features are used with classification methods, such as random forest (RF), classification trees (CT), support vector machines (SVM) and Naïve Bayes (NB), to classify bird sounds. Results indicated that the RF and CT models with all statistical features provide high accuracies at 0.9960 and 0.9988, respectively, reassuring the audience about the reliability of the research. The Boruta analysis, the mean and standard deviation of MFCC, and the number of samples were selected as important features for classification. After downscaling the number of features, the accuracy is 0.9958 for RF and 0.9544 for CT. Kappa statistics are higher for the above classification techniques. Support Vector Machines and NB classifiers showed low accuracies compared to RF and CT, suggesting the RF and CT models would be more appropriate for classifying bird sounds. The results obtained using machine learning techniques for statistical features demonstrate the potential for automated and accurate species identification. This approach enhances the ability to monitor avian biodiversity and opens new directions for ecological research and conservation strategies.
  • Thumbnail Image
    Item
    Determining hospital stay duration for sick children: a case study
    (Postgraduate Institute of Science (PGIS), University of Peradeniya, Sri Lanka, 2024-11-01) Gunawardhana, G. M. M. S.; Sumanarathna, M. A. A.; Abeysundara, S. P.
    Childhood diseases are a significant concern in the world, affecting the development and well-being of young children. The main objective of the study was to assess the length of stay of paediatric hospitalization of sick children admitted to the Sirimavo Bandaranayake Specialized Children’s Hospital (SBSCH) in Peradeniya, Sri Lanka. The dataset consisted of 104,757 paediatric patient admissions from 2019 to 2023, including the admission information such as admission number, age, gender, International Classification of Diseases code (ICD), date of admission and date of discharge. Among the total number of admissions, 58% were male patients. The preliminary results revealed that the number of patient admissions from 2020 to 2021 was significantly lower than in the other years. This could be due to the COVID-19 pandemic in Sri Lanka, which restricted people from gathering in public. However, the percentage of patient admissions per year with respect to gender was approximately the same for the period of study. Children under one year of age and school-aged children were the highest admitted patient category, reporting approximately 30% in each year. The most common disease type among children admitted was respiratory system diseases (14.96%), while other significant disease categories were injuries, infectious, factors influencing health status and digestive system problems. The average and the standard deviation of the length of stay in the hospital per admission were 3.06 days and 4.37 days, respectively. Mean comparison tests indicated that the length of hospital stay was statistically significant with respect to gender (3.11 days for males and 3.02 days for females) and the type of disease. A zero-truncated negative binomial regression model indicated that Age, Gender and Disease type were the most significant variables in determining the length of stay in the hospital. A chi-squared test for deviance indicated that the fitted model is significant compared to the null model (p < 0.01). The log count for the length of stay in days increases by 0.603 units for perinatal diseases, where it decreases by 1.457 for mental, behavioural and neurodevelopmental disorders. The log count for the length of stay decreases by 0.014 units for each one-year increase in age. The study provides information about the factors influencing the length of hospital stay for paediatric patients in SBSCH in Peradeniya, which could help healthcare providers optimize resource allocation and improve patient care outcomes.
  • Thumbnail Image
    Item
    Modelling record-breaking ODI cricket batting performances with extreme value theory
    (Postgraduate Institute of Science (PGIS), University of Peradeniya, Sri Lanka, 2024-11-01) Kumarage, K. D. Y. R.; Abeysundara, S. P.
    This study aimed to explore the highest individual scores recorded in One Day International (ODI) cricket matches through extreme value theory which further uncovers an area that has not been extensively studied in previous research. Data for this study was downloaded from ESPNcricinfo, covering the period from 1971 to 2024. The dataset comprised the career highest scores of 1,171 cricket players across eight countries with 24 variables, including player names, country, matches played, highest scores, strike rates, and other relevant metrics. Preliminary analysis reveals that the highest individual score recorded is 264, while the lowest is 0, with an average highest score of 60.85. It was observed that top-order and opening batters tend to have the highest ODI scores compared to players in other positions. Additionally, a multiple linear regression analysis identifies factors such as the number of matches played, batting average, and the number of fours and sixes have a significant effect on achieving a higher score throughout the career of a player. Also, players with longer careers tend to have the highest individual scores. Furthermore, extreme value theory was used to model the career highest scores recorded by individual players. Among the widely known extreme value distributions, it was revealed that the Weibull distribution is the most appropriate extreme value distribution for modelling. The estimated Weibull distribution parameters (scale = 1.00, shape = 60.85) indicate that the career highest scores are distributed with a heavy tail with a broader spread out to cover a wide range of highest scores. Furthermore, Weibull distribution is fitted for each country, and it is found that the distribution of the career highest scores for players from England is significantly different from other countries. The findings of this study will be valuable for sports analysts and coaches in understanding and optimizing extreme batting performances in cricket.
  • Thumbnail Image
    Item
    Spatial and temporal patterns of communicable diseases in Sri Lanka
    (Postgraduate Institute of Science (PGIS), University of Peradeniya, Sri Lanka, 2024-11-01) Aqeelah, M. S. F.; Abeysundara, S. P.
    Communicable diseases caused by bacteria, viruses, and parasites present major global health challenges. Understanding their transmission patterns is key to effective prevention and control. In Sri Lanka, limited research has been conducted to investigate the spatial and temporal patterns. The present study considered eight communicable diseases, including dengue fever, dysentery, encephalitis, enteric fever, leptospirosis, typhus fever, viral hepatitis, and food poisoning. Weekly data were collected from the Sri Lanka Epidemiology Unit website from 2007 to 2023. The current study found strong positive correlations among enteric fever, dysentery, and viral hepatitis. After 2015, dengue cases increased significantly, modelled by ARIMA(1,1,0)(0,0,1)[52], with a peak in 2017. Encephalitis cases rose from 2017 to 2023, best forecasted by ARIMA(1,1,1). Enteric fever declined after 2020, with ARIMA(1,1,2)(1,0,0)[52] indicating stability. Models were selected using AIC and BIC with residual diagnostics confirming forecast accuracy. Model validation showed predicted values within confidence intervals. Time series analysis also modelled typhus fever, viral hepatitis, food poisoning, and dysentery. Disease patterns shifted before and after COVID-19, potentially due to immunity changes caused by the pandemic. Dynamic time warping identified six clusters for dengue and four for leptospirosis, with K-medoid clustering showing better separation, supported by higher mean silhouette scores. Other diseases had less defined clusters, indicated by negative silhouette scores for both K-medoids and hierarchical clustering. This underscores challenges in accurate grouping due to small case numbers and the need for adaptive public health strategies. By identifying these patterns, the study informs targeted public health interventions, such as optimizing resource allocation, improving disease surveillance, and tailoring prevention strategies to specific regional and temporal trends in Sri Lanka.
  • Thumbnail Image
    Item
    Statistical analysis on child injury admissions: a case study
    (Postgraduate Institute of Science (PGIS), University of Peradeniya, Sri Lanka, 2024-11-01) Weerakoon, W. M. A. S.; Sumanarathna, M. A. A.; Abeysundara, S. P.
    Limited studies have been conducted in Sri Lanka on child injuries with respect to different types of injuries. The main objective of this study was to analyse the patterns in all types of injuries among newborn to 18-year-old children. The dataset was obtained from Sirimavo Bandaranayake Specialized Children’s Hospital, Peradeniya, Sri Lanka for the period 2019-2023. Records of 10,317 on child injury admission along with 18 variables including date of admission and discharge, age, gender, mode of discharge and information about the time, place and mechanism of injury. The preliminary analysis revealed that the average stay per admission is 1.52 days. The maximum length of stay during the period was 62 days. A time series plot revealed that the number of child injury admissions was at its lowest during the COVID-19 pandemic lockdowns. The highest number of daily injury admissions were reported on Saturdays, while during the pandemic period, it was on Fridays. Head injuries are the most common, and 79% of them were superficial injuries. Upper limbs were the second most injured body part, and more than half of them were fractures. The average length of stay in the hospital due to head injury (1.30 days) was significantly different from the average length of stay in hospital due to upper limb injuries (1.53 days). Association analysis indicated that there are statistically significant associations among the mechanism of injury, place of occurrence of injury, activity done at the time of injury, affected body region, nature of injury and age category. It was identified that age, time of injury, mechanism of injury, activity done at the time of injury, affected body region and nature of injury are significant in determining length of stay. The findings of this study lead healthcare professionals to look at specific injury types and associated variables, which will lead to the development of more efficient treatment practices.