Information Technology Journal KMUTNB

The Prediction of Bachelor Admission Trends in the University Using Data Mining Techniques

Sutarat Chaonafang — Sun, 05 Jan 2025 00:00:00 +0700

The declining number of students has become a pressing issue, impacting many universities worldwide, both in the present and the future. Currently, there is an increase in the number of public and private universities. However, university enrollment rates are decreasing, leading to intensified competition among universities for student admissions. The objective of this research is to predict fields of study trends in Bachelor degree of University applicants and to use this information for decision-making, resource allocation, and university curriculum development by using Data Mining Technique. The predictive models in this study comprise 3 formats: Naïve Bayes, Logistic Regression and Decision Tree. The dataset used in this research is sourced from the information system of the Academic Promotion and Registration Division at Nakhon Pathom Rajabhat University, encompassing undergraduate student records from the academic years 2014 to 2018, totaling 89,847 records. The results of comparing the performance of data classification models using K-Fold Cross Validation revealed that the best-performing prototype model is Logistic Regression, with an accuracy of 61.3%. This research can be applied to predict trends in prospective undergraduate students' selection of academic majors, inform strategic decision-making, and facilitate resource allocation within the university.

The Development of Management System for Connecting Color Scheme of Natural Indigo Dye with Indigo Dye Fabric Production

Kannikar Kamolrat, Akkarapon Phikulsri — Mon, 06 Jan 2025 00:00:00 +0700

This research involves the development of a management system for connecting the color scheme of natural indigo dye with indigo dye fabric production. The objectives are: 1) To develop a management system for connecting the color scheme of natural indigo dye with indigo dye fabric production. 2) To assess the efficiency of the management system for connecting the color scheme of natural indigo dye with indigo dye fabric; and 3) To assess the user satisfaction of the management system for connecting the color scheme of natural indigo dye with indigo dye fabric.

The research methodologies are as follows: study and collect data, analyze and design the system, and develop and test the system. The development of this system utilizes the Three-Tier architecture, consisting of: 1) The presentation layer, which uses Next.js in conjunction with Redux Toolkit, Material UI and React Leaflet. 2) The application layer, which employs Nest.js in combination with Sequelize ORM; and 3) The data layer, which utilizes the PostgreSQL database.

The results of the system development are as follows: 1) The management system for connecting the color scheme of natural indigo dye with indigo textile production has been successfully developed to manage three main parts: First, the general user section; Second, the producer or retailer section; Third the administrator section. 2) The evaluation of system efficiency conducted by five experts; they found it to be highly efficient, with a mean score of 4.84 and a standard deviation of 0.34.; and 3) The satisfaction evaluation, based on a sample of 50 individuals comprising customers and interested parties, as well as a group of producers or indigo-dyed fabric shops. The assessment revealed a very good level of overall satisfaction, with a mean score of 4.57 and a standard deviation of 0.58.

Sentiment Analysis from Tweets for Depression Level Prediction

Thara Angskun, Suda Tipprasert, Nantapong Keandoungchun, Jitimon Angskun — Sun, 05 Jan 2025 00:00:00 +0700

Currently, Thai people are increasingly suffering from depression, and these patients often do not know that they are depressed and often express themselves through social media because it is a form of communication through channels that do not rely on facial expressions. Therefore, this research presents sentiment analysis from Twitter users' tweets to predict their level of depression. Tweets used in the study include text, emoticons, and images. Sentiment analysis of those tweets applies hybrid machine learning, a combination of recursive feature selection using support vector machine and random forest modeling. The experimental results indicated that the developed provided the highest efficiency. The most important feature for predicting depression levels was the tweet's text type.

Forecasting Graduation By Data Mining Techniques

Kan Siraphonthanarat, Chutiphon Srisawat — Sun, 05 Jan 2025 00:00:00 +0700

This study intends to: 1) investigate the selection of significant features for data analysis; 2) create a predictive model for students' academic success in Pibulsongkram Rajabhat University's Faculty of Science and Technology; and 3) use data mining techniques to evaluate the model's efficacy. The 1, 082 records with 30 features that made up the data evaluated in this study were obtained from Rajabhat Pibulsongkram University's educational services department during the academic years 2560-2562. The study used information gain, and Chi Squared as feature selection methods in the analysis to determine the elements influencing academic achievement. According to certain parameters, each strategy entailed lowering variables with low weights. There were fifteen sets of data in the dataset. The data were split into two categories for model creation: training data (80%) and test data (20%). The model's performance was assessed by 10-Fold Cross Validation using data mining techniques, specifically Decision Tree, Random Forest, and NaÏve Bayes. Accuracy and F1-Score were the evaluation criteria. According to experimental results, the Random Forest model performed with the best overall accuracy when Information Gain values from dataset IG5 were matched with it.
The accuracy was 96.03%, and the average F1-Score was 88.65%.

An Automated Cyber Intrusion Prediction Model Using Deep Learning to Resilient in Cyber Threat for the Royal Thai Air Force

Somboon Udnan, Prasong Praneetplograng, Payap Sirinam — Sun, 05 Jan 2025 00:00:00 +0700

The Royal Thai Air Force was one of the nation’s Critical Information Infrastructure (CII) Organizations and has a record of cyber intrusions continue throughout the year. Therefore, researchers presented a new automated cyber intrusion prediction model using deep learning to resilient in cyber threat for the Royal Thai Air Force. It was an extension of the Looking Back Algorithm to increase the accuracy of the predictive model. In order to predict the future of the Air Force’s cyber threat patterns, researchers used cyber intrusion datasets from the Air Force that ranging from January 2021 to December 2021 with a total of 241,148 entries. We applied techniques such as RNN, LSTM, GRU, Bi-LSTM Deep Learning (DL). We developed the new cyber intrusion prediction model with name Bi-LSTM Looking Back Risk: Bi-LSTM-LBR. However, the developed model had high accurate result on test dataset that compared to other predictive models. In addition, prediction results had a Mean Absolute Error (MAE) was 0.038, a Mean Square Error (MSE) was 0.010 and a Root Mean Square Error (RMSE) was at 0.102.

Applying Semantic Matching and Annoy Index Methods to Analyze the Alignment of Courses in Curricula Designed to Meet the Future Workforce Competencies for 12 Target Industries within the Eastern Economic Corridor: A Case Study of Burapha University

Hemmarat Wachirahatthapong, Worawit Werapan — Sun, 05 Jan 2025 00:00:00 +0700

This research analyzes the alignment of Burapha University's curricula and courses in future workforce development guidelines for the 12 target industries within the Eastern Economic Corridor. It utilizes data from three sources, which include essential skill data for Eastern Economic Corridor organizations obtained from 98 organizations, and desired skills following the future workforce development guidelines for the 12 target industries. The data from both sources is consolidated to reduce redundancy, resulting in a total of 294 skills and curriculum data comprising 223 curricula and subjects within the university's curricula, totaling 10,650 courses. The Annoy Index for rapid data retrieval and semantic matching techniques is employed to analyze the semantic alignment of meaningful data between the desired skills from organizations and the curricula and courses. The analysis results, each with a similarity score of 0.66 for every target industry, offer valuable insights for planning curriculum development in alignment with the future workforce development guidelines for the 12 target industries. This research offers three policy-oriented recommendations for the strategic planning of curricula and course development of Burapha University to align with the future workforce development guidelines for the 12 target industries.

Developing a Chatbot to Promote Tourism in the Artisan Community in Nakhon Pathom Province

Sun, 05 Jan 2025 00:00:00 +0700

The objectives of this research are to 1) develop an information system to promote tourism through the Chatbot application system in tourist attractions of the artisan community in Nakhon Pathom Province 2) study tourists' satisfaction with the Chatbot with Python language to promote tourism through the application system. LINE Chatbot in Nakhon Pathom Province The data was collected from a sample of 423 tourists and members of online social groups, which is a specific sample group. Statistics used in data analysis include frequency, percentage, mean, and standard deviation. The research results found that the study of tourist behavior in using chatbots in 4 areas: meeting needs, being able to work. according to function, ease of use, and efficiency It was found that tourists' satisfaction in using chatbots was at a high level, with an average of 4.20. The aspect that tourists are most satisfied with is the ease of use. has an average of 4.23 in terms of meeting the needs and system performance They have an average of 4.20 and the ability to work according to their duties. has an average of 4.18, respectively.

Improving Performance of Convolutional Neural Network Models for Brain Tumor Classification from MRI Images through Hyperparameter Tuning

Pongsathorn Chedsom — Sun, 05 Jan 2025 00:00:00 +0700

Brain tumors are a common and severe disease, requiring timely diagnosis and treatment to reduce mortality rates and improve patient survival chances. This research presents a method for classifying brain tumors using deep neural networks. The Br35H dataset, consisting of 3,000 images (1,500 diseased and 1,500 non-diseased), was used for model training, divided into 70% training, 20% validation, and 10% testing sets. Additionally, 253 Brain MRI Images and 4,600 Brian Tumor images were used to test the model's accuracy. The experiments were divided into three groups: Group 1 employed popular architectures from existing literature without modification, using default hyperparameter settings. Specifically, the configurations included the Adam optimizer with a learning rate of 0.001 and a batch size of 32. The top five architectures with the highest accuracy were selected for further experiments. The architectures DenseNet201, Xception, InceptionResNetV2, MobileNetV2, and NasNetMobile accuracy of 98.00%, 98.00%, 98.00%, 97.67%, and 97.33% and loss of 0.08, 0.06, 0.08, 0.11, and 0.11 respectively. Group 2: Tuning three hyperparameters separately: Batch size, Optimizer, and Learning rate. The best results were Xception (Batch size 16) accuracy of 98.34% and loss of 0.08, InceptionResNetV2 (Learning rate 0.01) accuracy of 98.67% and loss of 0.13, and Xception (Optimizer Adamax) accuracy of 98.67% and loss of 0.06. Group 3: Tuning pairs of hyperparameters: The best architecture was InceptionResNetV2 (Batch size 32, Optimizer Adam, Learning rate 0.01) accuracy of 100.00% and loss of 0.001.

Life and Assets Safety Management of Communities with Technology and Innovation

Sun, 05 Jan 2025 00:00:00 +0700

This research endeavors to develop a system for the detection of knives and guns. Additionally, it aims to formulate guidelines to enhance the safety of life and property within the communities of Salaya Subdistrict Municipality, Phutthamonthon District, Nakhon Pathom Province. The process comprises two distinct phases: first, the development of a warning system that triggers when cameras detect knives and guns using the YOLOv8n model and second, conducting focus group to formulate guidelines aimed at enhancing the safety of life and property within the community. The research findings indicate that the system accurately detects knives and guns, achieving values of mAP@.5 = 0.829, precision = 0.836, and recall = 0.798. Additionally, real-time notifications are sent via Line Notify when weapons are detected. The formulated guidelines encompass three essential components: upstream process for prevention before an incident occurs in the middle of the water to suppress or stop the incident and downstream to help provide relief to those affected in various areas.

Establishing Chatbots Utilizing the Random Forest Classification for Division of Registration and Education Statistics for the Office of Academic Promotion and Registration: Rajamangala University of Technology Tawan-ok

Chumpol Mokarat, Duangjai Noolek — Sun, 05 Jan 2025 00:00:00 +0700

The paper presents the establishing chatbots utilizing the random forest classification for division of registration and education statistics for the Office of Academic Promotion and Registration for Rajamangala University of Technology Tawan-ok. The aim is to research best practices to implement an automated response system for education statistics and registration through online platforms in university departments, assessing the effectiveness of the system as well as the user’s acceptance and extent to satisfaction regarding the information system's use. Using the Random Forest Classification and the Python programming language, it constructed a model using the learning data set, which it then applied to the development of chatbots on the Dialogflow platform. The model evaluation resulted in an Accuracy of 97.47, Precision of 92.19, Recall of 93.67, and F-Measure of 92.92, the expert's assessment of efficiency had an average of 4.26, and the user group's satisfaction evaluation had an average of 4.28. To efficiently operate the organization's informative and news release services for educators, students, and related staff.