1 October 2025 in Journal of Mathematics and Science Teacher
This research aimed to investigate the relationship between mathematical proficiency and learning agility among high school students in Southern Thailand. The research focused on grade 10 students from two secondary schools in Surat Thani Province during the academic years 2021-2023, with a total sample of 363 students selected through purposive sampling. Mathematical proficiency was assessed through students’ academic performance in mathematics, while learning agility was measured using Hallenbeck’s (2016) framework, comprising four dimensions: seeking, sensemaking, internalizing, and applying using mathematics test. For data analysis, the research employed quantitative methods, including mean, standard deviation, independent samples t-test, and Pearson’s correlation coefficient. The study revealed that students generally demonstrated high levels of learning agility. No significant differences in learning agility were found between the two schools. However, students’ learning agility varied significantly based on their academic achievement (p < 0.05). Furthermore, a moderate, statistically significant positive correlation (r = 0.422; p = 0.002) was found between learning agility of applying dimensions and mathematical proficiency. These findings suggest that students with higher mathematical proficiency tend to exhibit stronger learning agility, enabling them to understand and apply mathematical concepts more effectively.
1 January 2024 in Journal of Southwest Jiaotong University
This article presents an application of data mining techniques and augmented reality (AR) technology in supporting an eCommerce system, focusing on three key data mining techniques: association rule mining, clustering, and forecasting. The FP-Growth algorithm has been used to analyze customer purchasing behaviors and discover relationships between products for product recommendations in the recommendation system (RS). In customer segmentation, RFM analysis was combined with K-means clustering to group customers based on purchasing behavior. Three types of forecasting models, Convolutional Neural Network (CNN), Stacked Long Short-Term Memory (Stacked LSTM), and Prophet, were compared to find the best sales forecasting model. The results showed that the association rule analysis led to the discovery of 6 rules. The resulting segmentation yields 4 customer segments with different behaviors: 'Important value', 'Important protection', 'General development', and 'Lost'. CNN emerged as the best-performing model in this study. In addition, AR has been used to create more immersive shopping experiences and attract customers more effectively. This article uniquely combines data mining techniques with AR to enhance eCommerce systems. It integrates FP-Growth for recommendations, RFM and K-means for customer segmentation, and compares CNN, Stacked LSTM, and Prophet for sales forecasting. The innovative use of AR to create immersive shopping experiences distinguishes this work from traditional studies.
13 March 2026 in Contemporary Mathematics and Science Education
Due to the problem of PISA test results showing mathematics scores below the average, along with teaching and learning issues where students still struggle with solving problems involving decimals then this research aims to achieve two primary objectives: (1) To compare students’ performance on decimal problems before and after implementing the bar model technique. (2) To examine the effectiveness of the bar model technique in enhancing students’ achievement in solving decimal problems, with a criterion of 70 out of 70. The sample for this study was specifically selected from a group of 29 students in the class. The students were then given a pre-test, and those who scored below 65% of the total score were selected. As a result, 15 students who struggled with solving decimal problems were identified. This research follows a mixed-methods approach, including qualitative data collection. Six students were randomly selected for interviews as the target group for qualitative analysis. The results revealed the following: (1) The post-test scores of the students were significantly higher than their pre-test scores at a significance level of 0.05. The post-test average score was 16.73, while the pre-test average score was 11.20. The discrepancy between the pre- and post-test scores was 5.53. (2) The effectiveness of the bar model technique on students’ achievement in solving decimal problems (E1 and E2) was found to be 79.77 percent and 83.65 percent, respectively, surpassing the specified threshold. Interviews with the target students indicated a clear increase in average scores. Students commented that the bar model improved their understanding of individual phrases, enabling them to analyze and represent them symbolically, leading to easier problem-solving. This research provides an alternative teaching method for mathematics. Recommendations for future research could explore the effects of the bar model on students with diverse learning styles or examine how students adapt this method across different cultural settings.