Admin Website UNIKOM

Tag: Garuda

  • Best Classifier Over Feature Selection and Delta Error in Time Series Dataset

    Best Classifier Over Feature Selection and Delta Error in Time Series Dataset

    Authors: Nugroho Widyanto, Jin-Whan Kim, Agus Nursikuwagus

    Abstract

    The selection of the best classifier on conventional machine learning is often made intuitively by observing existing research. It becomes an obstacle when stating high accuracy, regardless of whether the machine suits the dataset. This research proposes a method to deal with the accuracy of using a learning machine so that the accuracy generated can be aligned with the usage of the dataset. This method uses the feature selection method combined with the machine learning classifier. SelectKbest and principal component analysis (PCA) methods combined with the classifier machine by counting confirmed MSE, MAE, R2, and delta errors can predict the appropriate machine learning for the OTHERS datasets. Delta error and R2 by SelectKBest and the PCA can see the Delta Error tendency of its classifier learning. The observed classifiers confirmed the best R2 values on linear and Bayesian regression. The R2 values of Bayesian and linear regression are 0.634 and 0.687, respectively. Average Delta Error of MAE of SelectiKbest < Average Delta Eror of MAE of PCA, 3.21×1015 < 1.37×1016. In future research, we found the challenge of the best classifier in Deep Learning and the method of selecting features appropriate for the time series dataset. © 2024 Taylor’s University. All rights reserved.

    Author keywords

    Chi-square; Classifier; Delta error; Principal component analysis; SelectKBest

    This article can be accessed at https://www.scopus.com/pages/publications/85196260915

  • Analysis of Verticulture as an Alternative Agricultural Method in IndonesiaAuthors: Dafa Rizky Fahreza, Arif Abdan Syakur, Saeful Anwar Oktariansah, Bobi Kurniawan, Rudhi Lesmana

    Abstract
    The purpose of this study is to find out how effective Verti culture can be implemented in Indonesia. This study used qualitative method with literature review as data collection method. The result of this research shows that Verti culture is an alternative agricultural method which probably can help with global lack of food problem and global warming. It is supported by the facts that Verti culture is an efficient, high weather-resistant, and environment friendly agricultural method. The analysis result on this study can be used later as consideration to mass implement Verti culture in Indonesia. © 2024 Taylor’s University. All rights reserved.

    Author keywords
    Farming; Global lack of food; Global warming; Verti culture

    This article can be accessed at https://www.scopus.com/pages/publications/85211044891

  • Implementation of GraphQL in the Dodo Kids Browser Application

    Implementation of GraphQL in the Dodo Kids Browser Application

    Authors: Adam Mukharil Bachtiar, Dian Dharmayanti, Dimas Miftahul Huda

    Abstract

    This study aims to implement GraphQL to address the issues of under-fetching and over-fetching in the Dodo application. Dodo Kids Browser (Dodo) is a cross-platform application that serves as a parental control tool. This application assists parents in managing and monitoring their children’s online activities. The initial testing revealed problems with the backend of the Dodo application, which utilizes the REST API. Specifically, under-fetching and over-fetching were identified as issues that negatively impact performance and the application development process. After a comprehensive review of the existing literature, implementing GraphQL can serve as a viable solution. This solution is primarily due to the functions of GraphQL as a query language, enabling clients to precisely determine the specific data requested from the server. A series of tests were conducted to assess this implementation’s effectiveness. These tests involved comparing the required data attributes specified by the client with the actual data attributes transmitted from the server. Additionally, tests were performed to evaluate the performance of the previous and current systems. Upon analysing the results of these tests, it was discovered that GraphQL effectively resolves the issues of under-fetching and over-fetching that commonly arise in the Dodo Kids Browser application. © 2024 Taylor’s University. All rights reserved.

    Author keywords

    Indonesian sign language system; Learning media; Learning mode; Question mode; Sign language

    This article can be accessed at https://www.scopus.com/pages/publications/85196324733

  • Sign Language Letter Learning Media

    Sign Language Letter Learning Media

    Authors: Myrna Dwi Rahmatya, Mochamad Fajar Wicaksono 

    Abstract

    This research aims to build a sign language learning media, especially alphabet letters. Users can not only learn to act out the sign code but also train and test their understanding through question mode. The system will guide the user to act out the letters in sign language in the learning mode. After that, the user can practice in the practice mode, which is by acting out the sign code and the system will provide voice feedback. Meanwhile, with question mode, users can test their understanding of letters in sign language. In question mode, users will be asked to act out certain signs as instructed. The sign language used in this learning media is the Indonesian Sign Language System (SIBI). In question mode, users will act out the sign language with their hands and will be captured by the camera. In its development, the method used is the waterfall method with the stages of requirements, design, implementation, and testing. At the end of the stage, testing is done using the black-box method. Testing is done on the practice mode and question mode. The test results show that the learning media can work as expected. © 2024 Taylor’s University. All rights reserved.

    Author keywords

    Indonesian sign language system; Learning media; Learning mode; Question mode; Sign language

    This article can be accessed at https://www.scopus.com/pages/publications/85196284506

  • Queue System Analysis Using Arrival Patterns and Customer Service Patterns for Fast Food Restaurants at Simpang Dago Bandung

    Queue System Analysis Using Arrival Patterns and Customer Service Patterns for Fast Food Restaurants at Simpang Dago Bandung

    Authors: Faysal Rafly, Muhammad Rafi Hafizh, Agung Rahmat Fardan, Suryatno Wiganepdo Soegoto, Rudhi Lesmana

    Abstract

    Queues are one of the problems that have a large enough influence for a fast-food restaurant, especially in the world of food and beverage. Queues occur because the high arrival rate is not matched by the maximum level of service. This study aims to determine the queuing model that occurs at the McDonald’s cashier and find out whether the queue model applied is efficient by comparing the results of observations and simulation results. The research method used is descriptive analysis with a quantitative approach to customers at McDonald’s restaurants with arrival times and service times every 20 minutes on both servers. The results showed that there are several things that affect the queuing system, namely arrival time, service time, number of servers and McDonald’s consumer eating hours. From this, we can simulate the addition of servers to reduce queues during peak hours, namely at lunch time. Of course, the optimal performance of the queuing system will provide satisfaction for consumers or buyers because it can reduce waiting time during the purchasing process. © School of Engineering, Taylor’s University.

    Author keywords

    Fast-food restaurant; McDonald’s cashier; Queues; Queuing model; Simulation results

    This article can be accessed at https://www.scopus.com/pages/publications/85197583948

  • Developing Security Privacy Program in Information System

    Developing Security Privacy Program in Information System

    Authors: Muhamad Afghany Haryatno, Yeffry Handoko Putra

    Abstract

    The amount of information that people generate on daily life have put their smart devices as active participants of the business that runs by numerous organizations. However, following the process the information brought numerous people to consider the safety level, as some personal information might be leaked during the process. In the last few years, there have been a lot of national and international regulations that oblige organizations to follow certain rules in their information security program. This is mainly because leakage of personal information might put the victim’s privacy in danger, resulting in identity theft and perhaps financial loss. For organizations this situation could led them loss trust, reputation, and loyalty from their customers. Therefore, ensuring privacy during the process, storing, and sharing of personal information is important. Although it’s clear that security and privacy are two different terms, upon implementation security often believed covers privacy too. This paper seeks to enhance information system security, by developing privacy program using privacy by design as method. The goal is to illustrate and present some insight of how privacy by design address privacy in information system security. © School of Engineering, Taylor’s University.

    Author keywords

    Information system security; Privacy by design; Privacy program; Security

    This article can be accessed at https://www.scopus.com/pages/publications/85197658546

  • Development of a Machine Learning Model for Multiple Membership Publication Documents

    Development of a Machine Learning Model for Multiple Membership Publication Documents

    authors: Apriani P. Purfini, Stephanie B. R Hersianie, Rio Yunanto

    Abstract

    This research aims to develop a learning model in the text mining process by modifying TFIDF. The methodology employed in this study consists of a literature review, data collection, data preprocessing, and document processing. Text mining is conducted using the title and keyword attributes in publication documents. The outcome of this research is a learning model with modified TFIDF to determine multiple memberships in publication documents. The modification of TFIDF weighting can group words according to their document field categories. In contrast, previous TFIDF methods could only group words per document, making it difficult to determine the field categories of publication documents. The results of this research are expected to provide a solution to achieve more accurate outcomes with the same data scheme. The modification of TFIDF weighting can group words according to their document field categories. These field categories consist of a collection of words from many documents that have characteristics in specific field categories. Longer documents tend to have higher TFIDF weights because they can contain more words. This condition can affect the comparison between long and short documents. TFIDF also does not consider the context surrounding words, so in some cases, context is crucial for understanding the meaning of words or phrases. © School of Engineering, Taylor’s University.

    Author keywords

    Multiple membership; Publication document; TFIDF

    This article can be accessed at https://www.scopus.com/pages/publications/85199474137

  • Server Security and Air Quality Monitoring System Using Raspberry Pi 4 and Telegram

    Server Security and Air Quality Monitoring System Using Raspberry Pi 4 and Telegram

    Authors: Mochamad Fajar Wicaksono, Myrna Dwi Rahmatya

    Abstract

    The purpose of this research is to develop a system that uses Raspberry Pi and IoT technologies to monitor server room conditions such as temperature, humidity, air quality, and access rights security. The contribution of this research is that it makes it easier for users to monitor the server room and take appropriate action when something goes wrong. The method utilized in this study is experimental. The output obtained is in the form of a web display and telegram notifications. The brain of this system is the Raspberry Pi 4. The input received is temperature, humidity, carbon monoxide levels, carbon dioxide, and dust particle values. The thing that causes the Raspberry Pi 4 to activate the buzzer and send notifications via Telegram is when the temperature value is more than 230C, the humidity is less than 45% or 60%, the measured carbon monoxide value is more than 50 ppm, or carbon dioxide is more than 800 ppm or the AQI value of dust particles is more than 100. Regarding access rights, users can only enter the server room using a registered RFID card. Access rights violations will be detected by PIR1, PIR2, and Ultrasonic sensors. If this happens, the Raspberry Pi will take a photo with the Pi Camera, activate the buzzer, and send it via Telegram. The entire system testing process is tested into two main parts: temperature, humidity, AQI testing, and security testing related to access rights according to the program scenario and initial objectives. Test results show that every part is functioning 100%. The overall test shows 100% success where the system can do actions according to program conditions and scenarios. © School of Engineering, Taylor’s University.

    Author keywords

    AQI; Monitoring; Raspberry Pi4; Security; Server room; Telegram

    This article can be accessed at https://www.scopus.com/pages/publications/85198972754

  • Benefits of Using E-Learning: Analysed from Neuropsychological Aspects

    Benefits of Using E-Learning: Analysed from Neuropsychological Aspects

    Authors: Edi Firdaus, Munir M, Puspo Dewi Dirgantari, Chairul Furqon, Raden Heryaningtias Ramadhianti

    Abstract

    This research describe the benefits of using e-learning terms of the user’s neuropsychological aspect. A survey method was employed as the research methodology in this study., where the process of using e-learning by users is first described, then a survey is carried out by filling out a questionnaire to 286 elearning users based on 4 indicators, namely optimism, innovativeness, insecurity and discomfort from the survey results. The results obtained were then studied further concerning neuropsychological aspects. The results that the use of elearning shows a high sense of optimism and growth of innovation among users in but is still declared neutral in the insecurity and discomfort aspects. Then, looking at it from the neuropsychology aspect, e-learning can improve the emotional performance process in the brain, where the sense of optimism and innovativeness in oneself can be categorized as high. Recommendations for further research include reviewing the use of e-learning on other aspects of neuropsychology such as cognitive strength, trust, motivation, etc. © School of Engineering, Taylor’s University.

    Author keywords

    Discomfort; E-learning; Innovativeness; Neuropsychology; Optimism

    This article can be accessed at https://www.scopus.com/pages/publications/85218769534

  • Integration of Wireless Sensor Networks with Wireless Mesh Networks in IoT-Based Patient Monitoring Systems

    Integration of Wireless Sensor Networks with Wireless Mesh Networks in IoT-Based Patient Monitoring Systems

    Authors: Irawan Afrianto, Rahmat Mainusar, Sufa Atin

    Abstract

    Optimizing the patient handling process is an important part of a community health service centre. The study aims to develop a patient monitoring system using a Wireless Sensor Network (WSN) approach based on the Internet of Things (IoT) and Wireless Mesh Network (WMN). The application of the wireless sensor network in this study is used to monitor patients wirelessly and with precision in a nurse’s room. The method used in the development of this system is using the prototype method. The system model is developed in the form of applications and hardware consisting of IoT and wireless mesh network infrastructure. The application that was built has the function of receiving data from IoT devices that are sent via a wireless mesh network. Meanwhile, the developed hardware functions to detect the level of infusion water and the nurse calling device wirelessly. Wireless mesh network infrastructure is used as a data traffic medium that can configure and organize itself (self-configured), to maintain connectivity. The results obtained show that the prototype of the patient monitoring system is functionally running well, can produce fast and accurate information, and produces IoT devices that work efficiently and have reliable network infrastructure, to provide convenience for patients and nurses to obtain optimal health services. © School of Engineering, Taylor’s University.

    Author keywords

    Internet of things; Patient monitoring system; Wireless mesh network; Wireless sensor network

    This article can be accessed at https://www.scopus.com/pages/publications/85204008008