Journal of Mathematical Sciences and Informatics

VISUALISATION TECHNIQUES TO DETECT ABNORMAL SEA SURFACE TEMPERATURE DATA IN THE INDIAN OCEAN

2026-05-05T12:44:58+08:00

Sea Surface Temperature (SST) is the temperature of the water near an ocean’s surface. It plays a critical role in the interaction between the Earth’s surface and its atmosphere. The factors affecting sea surface temperature are crucial to understanding climate change, while SST itself is influenced by climatic factors such as humidity, air temperature, wind speed, and radiation. In practice, multivariate, anomalous abnormal data cannot be avoided. SST data is no exception to these data inconsistencies and may have multivariate or abnormal data. Therefore, a technique which visualises the abnormal data is crucial. In this study, Chernoff faces, and distance-distance plots are used as visualisation techniques to detect abnormal sea surface temperature data. Chernoff faces is a graphical representation of multivariate data, while a distance-distance plot is a graphical representation to reveal abnormal data. Based on both analyses, the abnormal data are clearly visible of climate factors of SST. Additionally, an Artificial Neural Network (ANN) with an autoencoder was used to detect outliers by applying a threshold to reconstruction errors. The Artificial Neural Network (ANN) model with the autoencoder and Chernoff faces method are effective techniques for detecting outliers in SST datasets.

AN APPLICATION OF BOUNDED AUTOCATALYTIC SET WITH SHANNON ENTROPY FOR DESKTOP COMPUTER SELECTION

2026-05-05T13:28:42+08:00

This article proposes a multi-criteria decision-making (MCDM) approach by integrating the Bounded Autocatalytic Set (BACS) method with Shannon entropy for desktop computer selection. Existing Graph Theory and Matrix Approach (GTMA)-based decision models often neglect the weighting of criteria, which may lead to less reliable ranking outcomes. To address this limitation, the proposed method incorporates Shannon entropy to objectively determine attribute weights before applying the BACS framework for ranking alternatives. The integration provides methodological advantages by combining a graphical representation of attribute interrelationships with entropy-based weighting, resulting in a more systematic and informative decision-making process. A numerical example involving five desktop computer models (M1–M5), evaluated across five attributes, is presented to demonstrate the applicability of the method. The results show that Model M3 achieves the highest permanent index value of 8.0396, indicating that it is the most preferred of the models being evaluated. The ranking results are further validated through comparison with the Graph Theory and Matrix Approach (GTMA) and a hybrid Analytic Hierarchy Process (AHP) – Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) method. Spearman correlation coefficients of 0.9 indicate a strong agreement between the proposed method and existing approaches, thereby confirming the reliability of the ranking outcomes. The findings demonstrate that the integration of BACS and Shannon entropy provides an effective and robust framework for solving MCDM problems.

PHYSICS-INFORMED NEURAL NETWORKS MODIFICATION FOR SOLVING 2D SHALLOW WATER EQUATIONS

2026-05-05T13:30:57+08:00

This study investigates the use of Physics-Informed Neural Networks (PINN) for solving the two-dimensional shallow water equations (SWE) on a flat bed and proposes a modified PINN with dynamic mesh-refinement strategy that adaptively increases the density of collocation points in critical or stiff regions such as propagating wave fronts and rapidly varying gradients. In the proposed framework, shallow-water physics is enforced as soft constraints in the loss, with initial and boundary conditions embedded to ensure a well-posed formulation without labelled targets. We evaluated fully connected architectures using the Adaptive Moment Estimation (Adam) and Limited-memory Broyden–Fletcher–Goldfarb–Shanno (L-BFGS) optimisation algorithms. Fully connected neural networks are trained using a combination of Adam and L-BFGS optimisers, and a FDM solution is employed as a reference for quantitative comparison. The standard PINN and its dynamically refined variant are evaluated against the FDM benchmark in terms of convergence behaviour and predictive accuracy. The results show that the refined PINN concentrates collocation points in stiff regions, reduces errors near wave fronts, and achieves accuracy closer to FDM with faster convergence.

HEAT TRANSFER OPTIMISATION FOR ROTATING TERNARY HYBRID NANOFLUID WITH OPPOSING MIXED CONVECTION PAST A VERTICAL FLAT PLATE

2026-05-05T13:35:49+08:00

The present study focuses on analysing and improving the heat -transfer performance of a rotating ternary hybrid nanofluid over a vertical flat surface under opposing mixed convection conditions. By employing suitable similarity transformations, the governing boundary-layer equations are reduced to nonlinear ordinary differential equations and subsequently solved using a MATLAB-based numerical approach. Furthermore, Response Surface Methodology (RSM) is used to investigate the combined influence of key parameters and to determine the conditions that maximise heat transfer efficiency. Increasing the concentration of nanoparticles, especially copper can greatly improve heat transfer efficiency, with copper nanoparticles showing the greatest enhancement, followed by aluminium oxide, (Al₂O3) and titanium dioxide (TiO₂) nanoparticles. Furthermore, desirability-based optimisation reveals that the heat-transfer rate attains a maximum value of 0.442152 with 99.93% desirability when the coded parameters A, B, and C (nanoparticle volume fractions) are at their maximum levels. Meanwhile, the decrement in skin friction along the x-direction is primarily influenced by the increase in volume fraction of copper nanoparticles, followed by titanium dioxide (TiO₂) and aluminium oxide nanoparticles. The findings provide significant insights into optimising heat-transfer in complex fluid dynamics systems, with potential applications in diverse industrial and engineering domains.

STATISTICAL ANALYSIS ON UNEMPLOYMENT RATE IN MALAYSIA

2025-12-02T14:32:52+08:00

Unemployment is a critical socio-economic issue affecting individual livelihoods and national economic performance. In Malaysia, the unemployment rate has fluctuated over the past decade, influenced by various social, economic, macroeconomic, demographic factors, and any other factors. The objectives of this research are to investigate the patterns and trends of the unemployment rate and to identify the significant factors that contribute to the joblessness in Malaysia. Several factors were considered such as the Gross Domestic Product (GDP), inflation, population growth, the COVID-19 pandemic, and Consumer Price Index (CPI). A correlation coefficient and multiple regression model were used to achieve the objective. Monthly data comprising 168 observations is collected from various sources such as the Department of Statistics Malaysia (DOSM) and the Central Bank of Malaysia. The dataset is categorised into two distinct periods, 2010-2019, representing the pre-pandemic era, and 2020-2023, which represents the COVID-19 pandemic era. The findings show that the CPI for food was both a consistent and significant predictor of unemployment across both periods under review, while the GDP was only significant before the pandemic, while the CPI for housing became a significant factor over the pandemic period. The remaining variables, such as inflation, the COVID-19 rate of infection, and population growth were not significant. In conclusion, the study showed that, while the GDP and CPI for food and housing have a significant effect on Malaysia’s unemployment rate, their impact changes between periods, which illustrates the shifting nature of economic determinants both before and after the COVID-19 pandemic.

TRIANGULAR INTUITIONISTIC FUZZY (TIF) TOPSIS AND STATISTICAL APPROACH FOR CONVENIENCE STORE PREFERENCE RANKING

2026-03-09T18:28:22+08:00

Convenience store selection often involves subjective judgement and uncertainty, making fuzzy decision-making approaches highly suitable. This study employs the Triangular Intuitionistic Fuzzy TOPSIS (TIF-TOPSIS) method, which integrates membership and non-membership functions to better capture uncertainty in consumer preferences. Five convenience stores anonymised as Store A, Store B, Store C, Store D, and Store E, were assessed based on four benefit criteria (cleanliness, store image, product assortment and service quality) and one cost criterion (price). Data was collected from undergraduate students enrolled in a mathematics degree programme at a public higher-learning institution in the Klang Valley. The TIF-TOPSIS procedure was implemented in two phases: (i) Conversion of linguistic evaluations into triangular intuitionistic fuzzy numbers, followed by aggregation, normalisation, and weighting; and (ii) computation of the Intuitionistic Fuzzy Positive Ideal Solution (IFPIS), Intuitionistic Fuzzy Negative Ideal Solution (IFNIS), and the closeness coefficient for ranking alternatives. Results indicate that Store C is the most preferred option, followed by Store B, Store D, Store A, and Store E. Cleanliness and service quality emerged as the most important criteria, while price was identified as the key cost factor. The fuzzy findings were supported by statistical analysis, which produced results that were both consistent with the ranking patterns and of acceptable reliability. The integration of TIF-TOPSIS with statistical validation enhances decision reliability and provides a comprehensive assessment framework for convenience store selection.

AN INTERACTIVE SPAM DETECTION MODEL USING AN ENSEMBLE ALGORITHIM

2025-07-30T13:07:28+08:00

Short Message Service (SMS) spam remains a significant issue in mobile communication systems. This study presents a framework for identifying spam messages in SMS communications using an ensemble of machine learning techniques for classification. The proposed framework involves a data preprocessing procedure to prepare the raw SMS dataset, followed by feature extraction using the Term Frequency–Inverse Document Frequency (TF-IDF) technique to represent textual data numerically. This enables the model to capture relevant characteristics from the processed data. During this experiment, the Ensemble model achieved 97% accuracy, outperforming Support Vector Machine (SVM) K-Nearest Neighbour (KNN), and Random Forest (RF), which achieved accuracies of 70.93%, 73.21%, and 79.18%, respectively. The proposed approach enhances user security in mobile communication and provides a comprehensive solution to the persistent issue of SMS spam, providing an effective spam detection system.

EMPIRICAL EVALUATION OF SPATIAL DYNAMICS OF SAND RESIDUE ON VARIOUS SHOE SOLE MATERIALS USING HARRIS CORNER STRENGTH FOR FORENSIC FOOTWEAR EXAMINATION

2026-02-05T17:12:50+08:00

Forensic footwear examinations often consider various types of geo-forensic materials that adhere to and or are retained by the soles of shoes. These particulates could potentially provide valuable insights regarding the movements of victims or suspects. Therefore, understanding the spatial dynamics of materials and the factors affecting the transfer, persistence, and recovery is essential for enhancing the reliability of geo-forensic investigations. This study aims to explore the factors affecting the spatial dynamics of residues on the shoe soles, and to determine the relative importance of different regions of the sole at retaining geo-forensic materials. Three volunteer test subjects of varying Body Mass Index (BMI) ratings were recruited for this study. Each of them walked through a sandy substrate using three types of worn footwear with soles made of rubber, Crosslite, and ethylene-vinyl acetate (EVA) foam. Then, the distribution of the sand particles beneath the soles was subsequently documented using digital photography. The captured images underwent preprocessing procedures including grayscale conversion, resizing, and normalisation. Each treated image was then segmented into 10 equal-sized non-overlapping regions, for detailed analysis using the Harris Corner detector. The total of the Harris corner response values within each region was used as a quantitative representation of residue retention, where higher corner strength values indicated greater accumulation of sand particles total. The findings revealed. that EVA foam had the highest retainability, followed by Crosslite and rubber. Meanwhile, the influence of the BMI statuses on the retainability was pronounced only on the rubber-soled shoes. Furthermore, the areas of the sole covering the metatarsal was found to be the most likely retain sandy particles. Overall, the effect of BMI statuses on the retainability of sandy particles was dependent on the material composition of the shoe sole, while the metatarsal region plays a dominant role in residue persistence. The results of this study add to the expanding pool of knowledge in forensic footwear analysis and lay the groundwork for the creation of increasingly complex analytical frameworks in accordance with the dynamic nature of residue transfer and retention in real-world investigations.