Students from Bangladesh pursuing STEM education often encounter obstacles when tackling diverse mathematical problems within various educational settings. Frequently, they find themselves lacking the essential prerequisite knowledge and strong foundational skills necessary to engage with the teaching and learning resources utilized at the undergraduate level, resulting in a significant number of students needing to seek readmission annually. The objective of this study is to explore the determinants of academic achievement among university undergraduates majoring in mathematics in Bangladesh. Employing a mixed-method research approach, the study combines quantitative and qualitative data analysis to examine the viewpoints of both students and educators concerning these factors. The authors primarily emphasize classifying the factors that impact the efficacy of mathematics pedagogical methods. The study is structured into three phases: i. An initial exploratory qualitative survey. ii. A quantitative triangulation survey. iii. Followed by explanatory semi-structured interviews. To begin, the initial qualitative survey identified significant factors that contribute to students' achievements and setbacks in mathematics. Subsequently, the quantitative analysis verified both similarities and distinctions in the perspectives of students and educators. Furthermore, the correlation coefficient analysis revealed that male students frequently exhibit inconsistency and a lack of enthusiasm for studying, resulting in subpar performance. Conversely, female students frequently cited challenges like the difficulty of connecting mathematical theories to real-world applications, heavy course loads, and limited resources as reasons for their academic difficulties. Lastly, insights from interviews with students highlighted their acknowledgment of inadequate study practices, excessive reliance on memorization, suboptimal teaching methods, low motivation, and external distractions as key factors leading to their struggles. They also recognized the importance of consistent practice, a solid comprehension of concepts, regular study routines, and effective learning strategies for successful mathematics education. In contrast, educators emphasized the significance of students having clear concepts, natural aptitude, motivation, and a sense of curiosity as pivotal elements for successful learning in mathematics. This conclusion suggests a new beginning in the realm of local mathematics pedagogy, achieved by scrutinizing teacher-student feedback about the factors influencing success and failure, considering the diverse individual and contextual variables at play. To foster mutual trust and understanding between students and teachers, it may be beneficial to engage in open discussions and interactions.

A Fuzzy Cognitive Map (FCM) is a soft computing method used to analyze and understand complex systems. This is particularly useful when there is a little i

Students’ cognitive function and participation in mathematical problems are adversely affected by mathematics anxiety. This research develops and evaluates a dynamic mathematical model (SAS: Susceptible-Anxious-Susceptible) to investigate the evolution and transmission of math anxiety within students over time. The model divides students into two subgroups based on presence and absence of math anxiety. Using epidemiological modeling concepts, we derive the basic reproduction number R₀, which determines the conditions for persistence (R₀ > 1) or elimination (R₀ < 1) of anxiety. Our analysis proves that the anxiety-free equilibrium is both locally and globally asymptotically stable when R₀ < 1, while the anxiety-prevailing equilibrium is stable when R₀ > 1, with the system undergoing a transcritical bifurcation at R₀ = 1. Sensitivity analysis using Partial Rank Correlation Coefficient (PRCC) reveals that the university entrance rate (π) and social transmission rate (β) positively impact R₀, while recovery rate (α) and exit rate (μ) negatively influence it. Notably, π and μ are identified as the most influential parameters. Numerical simulations demonstrate that the population of students with mathematics anxiety is highly sensitive to changes in β and μ, while remaining relatively stable with fluctuations in π and α. These findings provide a quantitative framework for developing effective interventions, suggesting that reducing social transmission of anxiety and enhancing recovery through supportive mechanisms can significantly curb math anxiety prevalence in educational settings.

Neural network (NN) models are widely used at present for forecasting infectious diseases. However, NN models are primarily data-driven, making them less effective when the available data are insufficient or noisy. Physics-informed neural network (PINN) models are special type of NN models that integrate both data and underlying physical laws governed by differential equations, making them more suitable for incorporation into mathematical models and improving forecasting accuracy. The aim of this study is to integrate a Susceptible-Vaccinated-Infected-Recovered (SVIR) model into three NN models, namely feedforward neural network (FNN), long short term memory (LSTM), and convolutional neural network (CNN) to forecast influenza cases in Mexico. In the first part, a deterministic SVIR model is formulated and standard mathematical analyses are performed, including the determination of equilibrium points and the basic reproduction number. In addition, local and global stability criteria are analyzed and existence of transcritical bifurcation phenomenon is examined near the disease free equilibrium. Sensitivity analysis reveals that the transmission rate and the recovery rate are the most influential parameters in controlling influenza propagation. In the second part, influenza case data spanning 120 weeks are used to train and test the three PINN models in conjunction with the SVIR differential equations. The performance of the models is evaluated using RMSE and MAE as error metrics. The results show that CNN model achieves the lowest RMSE and MAE values in both training and testing phases, demonstrating superior performance compared to the other two models. Finally, the SVIR deterministic model and all three SVIR-PINN models are used to generate forecasts for an additional 10 weeks. The findings of this study can assist the policymakers make better decisions to control the spread of influenza.