Journal of Science Ladkrabang

Concept to Practice: Guideline for Choosing between Covariance-Based SEM and Variance-Based SEM (Partial Least Square: PLS-SEM) in Structural Equation Modeling

2026-01-16T13:25:15+07:00

Structural Equation Modeling (SEM) is a quantitative analysis technique that is utilized to examine the relationship of factors to observed and unobserved variables (latent variable). Two major ways of conducting SEM are Covariance-Based SEM (CB-SEM) and Variance-Based SEM (VB-SEM), frequently referred to as Partial Least Squares (PLS). Both the approaches are based on different concepts and utilize different steps in the analysis process. This article conducts CB-SEM, through the AMOS program, and VB-SEM (PLS-SEM), through SmartPLS, using a case study of the Technology Acceptance Model (TAM), with a raw dataset provided from www.smartpls.com. The sample size is 1,190 and randomly 100 for the student version. In this case study, TAM was adapted for technical reasons with the same data and the same model. The findings reveal that CB-SEM is suitable for testing theoretical frameworks with a large dataset. In contrast, VB-SEM (PLS-SEM) is more suitable in cases with low samples or complex models. In conclusion, the article reveals the benefits and drawbacks of each methodology and provides some practical hints to help the researcher make the best choice for their work.

Modification of Coconut Coir Fibers to Enhance the Adsorption Efficiency of Methylene Blue Dye

2026-01-07T09:08:16+07:00

This study investigated the extraction and modification of cellulose from coconut coir fibers using hydrogen peroxide and sodium hydroxide for application as an adsorbent for methylene blue dye. After the modification process, the cellulose yield was 70%. The effectiveness of the modification process was confirmed through the analysis of functional groups, surface morphology, and elemental composition using Fourier Transform Infrared Spectroscopy (FTIR), Scanning Electron Microscopy (SEM), and Energy Dispersive Spectroscopy (EDS), respectively. The results indicated that lignin and hemicellulose were effectively removed, resulting in a rougher and more porous cellulose surface. The adsorption study revealed that the optimal conditions for methylene blue removal were a pH of 7, an initial dye concentration of 150 mg/L, an adsorbent dosage of 0.05 g, a contact time of 60 min, and a temperature of 30 °C, under which an adsorption capacity as high as 73.48 mg/g was achieved. The adsorption isotherm fitted well with the Langmuir model, indicating monolayer adsorption, while the adsorption kinetics followed the pseudo-second-order model, suggesting that the process occurred via a chemisorption mechanism. These results demonstrate that cellulose derived from coconut coir fibers possesses strong potential as an environmentally friendly adsorbent for the treatment of wastewater contaminated with organic dyes.

Dental Posture Detection Program for Dentists

2025-05-06T14:41:54+07:00

Dentists are often required to maintain the same posture for extended periods, which can lead to injuries if the posture is incorrect. To address this issue, a posture detection program for dentists has been developed using eight SparkFun 9DoF Razor IMU M0 sensors installed on key areas of the body, such as the head, shoulders, upper arms, and upper back. The system records movement data for both correct postures, such as sitting upright, and incorrect postures, such as leaning forward, which can cause muscle pain. The data is measured using Gyroscope sensors and transmitted via Wi-Fi to a program developed with Unity and C#. This program displays the dentist’s movements through a 3D model and shows motion angles in real-time. Testing results indicate that the system accurately records posture data and has received an average satisfaction score of 4.375 for wearability and 4.125 for display performance, out of a maximum score of 5.

Effects of Missing Data Patterns and Rates on the Accuracy of ARIMA Model for Electricity Consumption Forecasting

2026-02-03T16:31:18+07:00

This study aims to investigate the effects of missing data patterns and missing rates on the forecasting accuracy of the ARIMA model from a methodological perspective using monthly electricity consumption data in Thailand from 2002 to 2025. A full factorial design was employed to simulate missing data under the Missing Completely at Random (MCAR) and Missing at Random (MAR) mechanisms at missing rates of 10%, 20%, and 30%, with 50 replications conducted for each experimental condition. Three widely used imputation methods in time series forecasting were applied, namely Last Observation Carried Forward (LOCF), Linear interpolation, and Kalman filtering. The results reveal that the imputation method is the most influential factor affecting forecasting accuracy, followed by the missing data rate. Kalman filtering consistently produced the lowest Root Mean Squared Error (RMSE) and demonstrated high stability across experimental conditions, whereas Linear interpolation consistently yielded the highest RMSE values. In addition, the performance of certain methods, particularly LOCF, varied substantially according to the proportion of missing data. These findings suggest that Kalman filtering is a robust and appropriate approach for handling incomplete energy time series data and can effectively support long-term energy consumption planning under data uncertainty.

Integrated Storage Location Assignment for E-Commerce Product Categories Using Product Affinity Analysis

2025-12-17T19:12:32+07:00

This study focuses on improving warehouse management efficiency for an e-commerce business. The primary issue identified in the case study of ABC Co., Ltd. was inefficient product storage, which resulted in excessive picking time and travel distance for warehouse operators. To address this problem, the study proposes an improved storage strategy based on two months of historical order data, integrating Class-based storage and Product affinity analysis. Products were classified into three main groups using ABC analysis to support inventory segmentation, with particular emphasis placed on Group A items due to their highest picking frequency. The baseline analysis revealed that the total picking distance prior to warehouse reorganization was 923,868.5 meters. By applying mathematical optimization models and solving them using Excel Solver, the Class-Based Storage approach reduced the total picking distance to 153,207 meters, while the Product Affinity Analysis model achieved a further reduction to 150,471 meters. Experimental results indicate that the Product Affinity Analysis approach, which considers frequently co-purchased items, enables more effective storage location assignment and achieves the highest improvement in picking efficiency, reducing total travel distance by up to 83.7%.

A Prototype System for Detecting Land Snails in Experimental Vegetable Plot Using YOLOv10-X Deep Learning Model

2025-04-03T16:23:10+07:00

This paper presents a prototype system for detecting land snails in an experimental vegetable plot using YOLOv10-X Deep learning model. In this proposed system, (Internet of Things) devices were used in conjunction with computer vision techniques and YOLOv10-X Deep Learning model. Users can view the results of land snail detection in real time, view land snail detection reports via mobile application, and receive land snail detection notifications via the LINE application. The land snail species used in this study were the African giant snail and the Siamese snail. From random testing of the detection accuracy in 100 video image frames, it was found that the proposed system gave an accuracy of 74.15% for detecting African giant snails, 64.08% for Siamese snails, and mean average precision (mAP) of both classes was 97.31%.

Comparative Performance of Poisson-Lindley Distributions Using Real Data Sets

2026-02-19T09:26:35+07:00

This study aims to compare the performance of three distributions of the Poisson-Lindley distribution, namely the generalized Poisson-Lindley distribution, the weighted Poisson-Lindley distribution, and the three-parameter Poisson-Lindley distribution. The comparative analysis is conducted against the Poisson and negative binomial distributions, which are commonly used as baseline models for count data. The study utilizes three real data sets: the number of road traffic accident fatalities in Rayong Province, the number of deaths from coronavirus disease 2019 in Sisaket province obtained from the open government data center of Thailand, and the number of physician visits from the Australian national health survey available in the AER package in the R program. Parameter estimation is performed using the maximum likelihood method. The goodness-of-fit of the models is evaluated using the -Log-likelihood value (-LL), Akaike's Information Criterion (AIC), Bayesian Information Criterion (BIC), and the Kolmogorov-Smirnov (KS) test statistic. The results show that the generalized Poisson-Lindley distribution provides the most suitable fit for three count data sets exhibiting overdispersion.

Building Makruk Endgame Tablebases: A Computational Study of Ma-Bia Ngai and Khon-Bia Ngai Endgames

2026-04-09T15:04:57+07:00

This research aims to construct Endgame tablebases for the Makruk (Thai Chess) game in order to analyze the outcomes of two complex endgame patterns 1) Ma-Bia Ngai (Knight with Promoted Pawn) and 2) Khon-Bia Ngai (Bishop with Promoted Pawn), considering whether these positions result in a win or a draw according to the rules of Makruk. In the first pattern, the chasing side consists of one King, one Knight, and one Promoted Pawn, while the escaping side has one King. Computer analysis revealed a total of 11,930,016 legal positions, of which the chasing side can force a win in 472,900 positions, requiring a maximum of 36 moves, accounting for 3.96% of all positions. The remaining 11,457,116 positions (96.04%) result in a draw. This pattern is not affected by the Makruk rule on counting the Knight’s moves, which allows the escaping side to count up to 64 moves. In the second pattern, the chasing side consists of one King, one Bishop, and one Promoted Pawn, while the escaping side has one King. Analysis found a total of 12,170,304 legal positions, of which the chasing side can force a win in 10,438,976 positions, requiring up to 57 moves, accounting for 85.77% of all positions, and 1,731,328 positions (14.23%) are draws. However, according to the Makruk rule on counting the Bishop’s moves, the escaping side may count up to 44 moves, which must be adjusted based on the total number of pieces on the board (4 pieces: one chasing King, one Bishop, one Promoted Pawn, and one escaping King). Consequently, the chasing side can only move for 41 moves, and if it fails to checkmate within this limit, the game is considered a draw. This adjustment reduces the number of positions where the chasing side can force a win to 10,160,936, lowering the winning ratio to 83.49%, as 278,040 positions require more than 41 moves to complete.

A Comparative Risk Analysis of Investment Portfolios Consisting of Stocks and Cryptocurrencies Using Value at Risk Method

2026-05-14T09:59:38+07:00

This research examines and compares the risk levels of investment portfolios composed of stocks and cryptocurrencies using the Value at Risk approach. The objective is to evaluate the risk profiles of both asset classes and compare Value at Risk (VaR) estimates calculated through three distinct methodologies: Historical Simulation, Variance-Covariance (Parametric), and Monte Carlo Simulation. The study utilizes daily price fluctuations of ten selected assets, comprising five stocks: BCP, DELTA, ICT, KTB, and TRUE and five cryptocurrencies: TrueUSD (TUSD), Binance Coin (BNB), Ethereum (ETH), Bitcoin (BTC), and Solana (SOL). The analysis is based on an equally weighted portfolio with a total value of 1,000,000 THB. The dataset spans a three-year period from September 1, 2021, to August 31, 2024. The empirical results reveal that at a 90% confidence level, the VaR estimates are 2.88%, 2.04%, and 2.12% for the Historical simulation, Variance–covariance, and Monte Carlo methods, respectively. At the 95% confidence level, the values increase to 4.50%, 3.17%, and 3.34%, while at the 99% confidence level, they reach 9.18%, 5.27%, and 5.39%, respectively. Ultimately, this study suggests that investors can apply VaR techniques to evaluate and compare risks in hybrid portfolios. Such an approach enables more informed decision-making aligned with individual risk tolerance and provides a more precise evaluation of potential financial exposure.

Sustainable Management of Fluorescent Lamp Waste in Local Waste Management Areas

2026-03-04T16:35:55+07:00

This study aimed to evaluate mercury contamination in the air within fluorescent lamp waste storage areas and to analyze the effect of ventilation on reducing airborne mercury concentration, in order to develop safe and sustainable waste management guidelines. The results revealed that rural areas still generate and use fluorescent lamps at significantly higher proportions than urban areas. Measurements of airborne mercury in storage facilities showed that closed areas exhibited mercury concentrations more than 50 times higher than open areas. Specifically, non-ventilated storage rooms recorded an average concentration of 520.9 ± 160.3 ng/m³, whereas open areas averaged only 10.9 ± 5.4 ng/m³. Controlled chamber tests indicated that broken fluorescent lamps released a large amount of mercury vapor during the first 10 minutes, with a total average emission of approximately 8 milligrams per lamp. Simulation using the mass balance model demonstrated that cross ventilation was significantly more effective in reducing mercury concentration than single-sided ventilation. When properly sized openings were applied, the mercury concentration decreased dramatically—from 520.8 ng/m³ to only 21.7 ng/m³—highlighting the direct correlation between ventilation rate and reduction of mercury accumulation. In conclusion, effective management of fluorescent lamp waste should begin with proper source separation between fluorescent and LED lamps, the use of durable and sealed containers during transport, and storage in facilities with adequate ventilation rates, all under a legally compliant hazardous waste management system. These measures are essential to prevent environmental contamination and mitigate long-term health risks in a sustainable manner.