Jurnal Politeknik Negeri Batam (PoliBatam)
Not a member yet
3001 research outputs found
Sort by
Desain Sistem Otomatisasi Kontrol Pompa Air dengan Dua Saluran Keluaran
No full textPompa air merupakan alat yang berfungsi untuk memindahkan air dari sumber menuju tempat yang diinginkan. Dalam penelitian ini, satu unit pompa air dimanfaatkan untuk memenuhi kebutuhan air dua rumah yang berdekatan dan memiliki konsumsi air yang hampir sama. Studi ini mengkaji performa sistem otomatisasi pengisian baterai yang mengandalkan pompa 12V, panel surya, serta baterai sebagai sumber energinya. Sistem bekerja dengan mentransmisikan informasi ketinggian air dari Node ke Gateway. Berdasarkan pengujian, rata-rata waktu yang dibutuhkan untuk mengisi tandon pertama adalah 36 menit, sementara tandon kedua memerlukan 44 menit. Jika kedua tandon diisi secara bersamaan, waktu pengisian menjadi 44 menit untuk tandon pertama dan 46 menit untuk tandon kedua. Pengisian daya baterai melalui panel surya membutuhkan waktu sekitar 6 jam saat cuaca cerah, dan pompa mampu beroperasi secara terus-menerus selama 12,5 jam menggunakan baterai 12V 50 Ah. Penelitian ini menyajikan evaluasi sistem serta memberikan wawasan untuk meningkatkan efisiensi penggunaan energi dalam proses pengisian air secara otomatis.Pompa air adalah perangkat yang digunakan untuk memompa air dari sumbernya ke lokasi yang diinginkan. Dalam kasus penggunaan satu pompa air untuk dua rumah, pompa ini berfungsi sebagai penyedia air bagi kedua rumah tersebut. Hal ini dapat diterapkan pada dua rumah yang berdekatan dan memiliki kebutuhan air yang relatif serupa. Penelitian ini mengevaluasi kinerja sistem otomatisasi pengisian baterai yang menggunakan Pompa Air 12V, Panel Surya, dan baterai sebagai sumber daya pompa. Sistem ini bertujuan sebagai penyedia air untuk dua rumah dengan menggunakan satu pompa air, terutama pada dua rumah yang berdekatan dengan kebutuhan air yang serupa. Alat ini beroperasi dengan cara mengirimkan data ketinggian air dari Node ke Gateway ketika mencapai 70 cm, yang kemudian mengaktifkan pompa air dan solenoid valve untuk mengisi tandon hingga ketinggian 30 cm. Setelah mencapai level tersebut, pompa air dan solenoid valve dimatikan. Hasil pengujian menunjukkan waktu rata-rata pengisian tandon adalah 36 menit untuk tandon 1 dan 44 menit untuk tandon 2. Ketika keduanya mengisi bersamaan, waktu rata-rata adalah 44 menit untuk tandon 1 dan 46 menit untuk tandon 2. Pengisian baterai menggunakan solar panel memerlukan waktu 6 jam pada cuaca matahari terik, dan pompa dapat beroperasi selama 12,5 jam secara terus-menerus dengan menggunakan baterai 12V 50Ah. Hasil dari analisa lainnya menunjukkan adanya delay pada sistem LoRa yang disebabkan oleh sumber daya yang tidak mencapai 3.3V, menyebabkan kinerja yang tidak efisien. Penelitian ini memberikan gambaran tentang kinerja sistem otomatisasi pengisian baterai dengan berbagai komponen dan kondisi, serta memberikan wawasan untuk meningkatkan efisiensi dan efektivitas penggunaan energi dalam pengisian air
A Study on Reducing Measurement Fluctuations in Iron-Electrode Salinity Sensor Using Moving Average and Kalman Filters
No full textIndonesia is recognized as one of the leading shrimp-producing countries globally, with most farms operating on a small scale using traditional methods. This creates a strong demand for low-cost technologies to support aquaculture. One critical component in shrimp farming is water quality monitoring, where salinity is a key parameter affecting shrimp health and growth. Affordable salinity sensors using iron electrodes are increasingly considered. However, they often produce unstable and fluctuating readings, compromising monitoring reliability. This study addresses the issue by applying digital filters to enhance the stability of salinity sensor data. Two filtering methods, Moving Average and Kalman filters were evaluated using salinity ADC data from previous research. The analysis focused on comparing their effectiveness in stabilizing measurements. Results show that the Moving Average filter outperformed the Kalman filter, providing lower standard deviation values (87,09, 65,69, 63,67) and variance values (7,5807E+03, 4,3150E+03, 4,0542E+03), confirming its suitability for improving low-cost salinity sensor performance
The Development of a Deployment System Architecture for a Flask-Based Chatbot Using an LSTM NLP Model for Customer Service Question & Answer
No full textIn the past two decades, the rapid growth of e-commerce has significantly transformed global business practices. E-commerce has not only revolutionized the retail industry but also positively impacted businesses and consumer experiences. The ease of online shopping enables users to select products at more competitive prices. Amidst these changes, human-computer interactions have increasingly evolved toward natural conversations through Natural Language Processing (NLP). This study aims to develop a chatbot utilizing Long Short-Term Memory (LSTM) technology as a medium for e-commerce customer service. The dataset used for chatbot development is in JSON format and consists of 580 entries spanning 38 categories or classes. Data processing involves several preprocessing stages, including case folding, lemmatization, tokenization, and padding. The model is developed using a bidirectional LSTM and GRU architecture, followed by regularization techniques to enhance performance. Evaluation results show the model achieves 90% training accuracy and 63% validation accuracy with an F1-score of 62%. While there are indications of overfitting, the observed differences are not statistically significant, indicating the model remains capable of providing reliable responses. Additionally, the model is integrated into a Flask-based web application with an interactive interface to facilitate user access. This study demonstrates that LSTM is effective in addressing vanishing gradient problems.In the past two decades, the rapid growth of e-commerce has significantly transformed global business practices. E-commerce has not only revolutionized the retail industry but also positively impacted businesses and consumer experiences. The ease of online shopping enables users to select products at more competitive prices. Amidst these changes, human-computer interactions have increasingly evolved toward natural conversations through Natural Language Processing (NLP). This study aims to develop a chatbot utilizing Long Short-Term Memory (LSTM) technology as a medium for e-commerce customer service. The dataset used for chatbot development is in JSON format and consists of 580 entries spanning 38 categories or classes. Data processing involves several preprocessing stages, including case folding, lemmatization, tokenization, and padding. The model is developed using a bidirectional LSTM and GRU architecture, followed by regularization techniques to enhance performance. Evaluation results show the model achieves 90% training accuracy and 63% validation accuracy with an F1-score of 62%. While there are indications of overfitting, the observed differences are not statistically significant, indicating the model remains capable of providing reliable responses. Additionally, the model is integrated into a Flask-based web application with an interactive interface to facilitate user access. This study demonstrates that LSTM is effective in addressing vanishing gradient problems
Sentiment Analysis of Youtube and Gotube Reviews on Google Play Using the Support Vector Machine (SVM) Method in Indonesia
No full textThis research, titled Sentiment Analysis of YouTube and GoTube Reviews on Google Play Using the Support Vector Machine (SVM) Method in Indonesia, analyzes user perceptions of YouTube and GoTube based on Google Play reviews. The study is motivated by the growing popularity of video streaming apps in Indonesia and the limited sentiment analysis research on these platforms. The research collects 1,600 reviews (800 per app) from 2023-2024 using Python’s Scrapy library. The data is split 70% for training and 30% for testing, undergoing text preprocessing (tokenization, stop word removal, stemming), TF-IDF weighting, and SVM classification with an RBF kernel. Evaluation metrics include accuracy, precision, recall, and F1-score, with PCA used for visualization. Results show 94.50% accuracy overall, 97.01% for YouTube, and 92.66% for GoTube. GoTube has higher positive sentiment (385 of 400 test reviews) than YouTube (345 of 400) but lower negative sentiment (15 vs. 55). However, the model exhibits a positive class bias due to data imbalance. The study concludes that SVM effectively detects positive sentiment, but balancing data and exploring non-linear methods could improve negative sentiment detection.This research, titled Sentiment Analysis of YouTube and GoTube Reviews on Google Play Using the Support Vector Machine (SVM) Method in Indonesia, analyzes user perceptions of YouTube and GoTube based on Google Play reviews. The study is motivated by the growing popularity of video streaming apps in Indonesia and the limited sentiment analysis research on these platforms. The research collects 1,600 reviews (800 per app) from 2023-2024 using Python’s Scrapy library. The data is split 70% for training and 30% for testing, undergoing text preprocessing (tokenization, stop word removal, stemming), TF-IDF weighting, and SVM classification with an RBF kernel. Evaluation metrics include accuracy, precision, recall, and F1-score, with PCA used for visualization. Results show 94.50% accuracy overall, 97.01% for YouTube, and 92.66% for GoTube. GoTube has higher positive sentiment (385 of 400 test reviews) than YouTube (345 of 400) but lower negative sentiment (15 vs. 55). However, the model exhibits a positive class bias due to data imbalance. The study concludes that SVM effectively detects positive sentiment, but balancing data and exploring non-linear methods could improve negative sentiment detection
Implementation of FP-Growth Algorithms for Promo Package Determination in a Scooter Motorcycle Workshop Business
No full textThis study applies the FP-Growth algorithm to design bundled promotions for a scooter motorcycle accessory store and workshop in Denpasar, Bali. FP-Growth was chosen for its efficiency in mining frequent itemsets without generating candidate sets. From 23,381 transaction records (January-August 2024), the algorithm identified 16 association rules using a minimum support of 1% and confidence of 50%. These rules were selected based on lift values and product relevance. One notable example is the association between "BAUT TITANIUM GR5 M10 X 60" and "BAUT TITANIUM GR5 M8X50", which had a lift of 47.814, indicating a very strong co-purchase relationship. These high-lift combinations present valuable opportunities for bundling and targeted point-of-sale offers. The algorithm performed efficiently, with a runtime of just 0.1354 seconds and 402.6 MB of memory usage. Bundles based on these associations were presented to customers, and feedback was collected through a Customer Satisfaction (CSAT) survey involving 56 recent buyers. The survey yielded a high CSAT score of 83.93%, demonstrating customer satisfaction with the bundles’ relevance and appeal. These results confirm that FP-Growth can effectively inform promotional strategies by identifying strong product pairings that align with actual purchasing behavior. Strategically promoting such bundles not only enhances customer experience but also encourages multi-item purchases. This data-driven bundling approach is practical and profitable for medium-sized retail businesses, ultimately supporting the goal of increasing the Average Order Value
Sentiment Analysis of User Reviews of the AdaKami Online Loan App from the App Store Using SVM and Naive Bayes
No full textThis study aims to classify sentiments on user reviews of the AdaKami online loan application, which are obtained through web scraping techniques from the Apple App Store platform. A total of 2000 reviews were collected, then selected and 1000 reviews were selected to be manually labeled by two linguistic experts, to ensure the validity of the classification. Sentiments are divided into three categories, namely negative, neutral, and positive. The classification model was built using two machine learning algorithms, namely Support Vector Machine (SVM) and Naïve Bayes (NB). The evaluation was carried out by measuring accuracy, precision, recall, F1-score, as well as through confusion matrix and cross-validation. The results showed that SVM performed better, with an accuracy of 97.5%, an F1-score of 0.97, and an average cross-validation accuracy of 84.69%. In contrast, Naïve Bayes recorded an accuracy of 81.4% and an F1-score of 0.77. The results of the paired t-test showed that the difference in performance between the two models was statistically significant (p < 0.05). The SVM model was then applied to predict 971 unlabeled reviews, and the results showed a dominance of negative sentiment. Wordcloud visualizations reinforced this finding, with words such as “bilih”, “bunganya”, and “teror” as the most frequently occurring words. These findings prove that SVM is more effective in classifying online loan review sentiments, as well as providing important insights for developers in understanding user perceptions and experiences.This study aims to classify sentiments on user reviews of the AdaKami online loan application, which are obtained through web scraping techniques from the Apple App Store platform. A total of 2000 reviews were collected, then selected and 1000 reviews were selected to be manually labeled by two linguistic experts, to ensure the validity of the classification. Sentiments are divided into three categories, namely negative, neutral, and positive. The classification model was built using two machine learning algorithms, namely Support Vector Machine (SVM) and Naïve Bayes (NB). The evaluation was carried out by measuring accuracy, precision, recall, F1-score, as well as through confusion matrix and cross-validation. The results showed that SVM performed better, with an accuracy of 97.5%, an F1-score of 0.97, and an average cross-validation accuracy of 84.69%. In contrast, Naïve Bayes recorded an accuracy of 81.4% and an F1-score of 0.77. The results of the paired t-test showed that the difference in performance between the two models was statistically significant (p < 0.05). The SVM model was then applied to predict 971 unlabeled reviews, and the results showed a dominance of negative sentiment. Wordcloud visualizations reinforced this finding, with words such as “bilih”, “bunganya”, and “teror” as the most frequently occurring words. These findings prove that SVM is more effective in classifying online loan review sentiments, as well as providing important insights for developers in understanding user perceptions and experiences
Analysis of Internet Service Provider Selection Using the Analytical Hierarchy Process (AHP) and Simple Additive Weighting (SAW) Methods
No full textCurrently, the need for an internet connection is increasing due to the human need for information and communication that can be obtained through the internet. A decision support system with criteria is needed to solve the challenge of choosing an internet package with many components or criteria that must be evaluated. Analytical Hierarchy Process (AHP) and Simple Additive Weighting (SAW) are two multi-criteria decision support system techniques, but they use different measurement techniques. This research aims to identify the best Internet Service Provider by applying the Simple Additive Weight (SAW) and Analytic Hierarchy Process (AHP) techniques. The objects used in this research are the providers Indihome, Biznet, and Solnet. The weight of each criterion is determined using the AHP technique, and the best suppliers are ranked using the SAW approach. The sampling technique used in this research was Purposive Sampling. In this study, the ranking results of 100 respondents showed that the ranking for the AHP method was 60% and for the SAW method was 92%. Thus, global calculations using the AHP and SAW methods were obtained for the best alternative ISP, namely Indihome
PENGEMBANGAN PROSES BISNIS DI PUSAT KEUNGGULAN UNTUK MENINGKATKAN OPTIMASI OPERASIONAL
No full textThis research aims to develop business processes at the Center of Accounting and Fintech (CoE Afintec) Polibatam. The analysis of this business process development is an effort to improve operational efficiency and support the achievement of the strategic goals of CoE Afintec. The research method used is a qualitative method with an applied research type, which is focused on solving real problems faced by CoE Afintec. Data were collected through in-depth interviews with related parties and then analyzed using a BPMN (Business Process Model and Notation)-based Business Process Diagram (BPD) to map core processes and see positive and negative values. The research results showed that several processes were not running effectively and optimally. Based on these results, a new business process was redesigned to be more structured, responsive, and in line with the needs and objectives of CoE Afintec development. With the development of this business process, it is hoped that CoE Afintec Polibatam can strengthen its role sustainably as a center of excellence that supports the Tridharma of Higher Education and becomes a profit center for the Polibatam Business Management Department.Penelitian ini bertujuan untuk mengembangkan proses bisnis pada Center of Accounting and Fintech (CoE Afintec) Polibatam. Analisis pengembangan proses bisnis ini sebagai upaya meningkatkan efisiensi operasional dan mendukung pencapaian tujuan strategis CoE Afintec. Metode penelitian yang digunakan adalah metode kualitatif dengan jenis penelitian terapan, yang difokuskan pada pemecahan masalah nyata yang dihadapi CoE Afintec. Data dikumpulkan melalui wawancara mendalam pihak terkait, kemudian dianalisis menggunakan Business Process Diagram (BPD) berbasis BPMN (Business Process Model and Notation) untuk memetakan proses-proses inti dan melihat nilai positif dan negatif. Hasil penelitian menunjukkan bahwa terdapat beberapa proses yang berjalan tidak efektif dan optimal. Berdasarkan hasil tersebut, dilakukan perancangan ulang proses bisnis baru yang lebih terstruktur, responsif, dan sejalan dengan kebutuhan dan tujuan pengembangan CoE Afintec. Dengan adanya pengembangan proses bisnis ini, diharapkan CoE Afintec Polibatam dapat memperkuat perannya secara berkelanjutan sebagai pusat keunggulan yang mendukung Tridharma Perguruan Tinggi dan menjadi profit center bagi jurusan Manajemen Bisnis Polibatam
Performance Comparison of Wheeled Soccer Robot Frames Using Finite Element Analysis (FEA)
No full textFrame analysis plays a vital role in the Indonesian Wheeled Soccer Robot Contest (KRSBI-B) by identifying design weaknesses and enabling optimization before building the physical robot. This study compares three frame designs: standard, modified standard, and the Tech United team\u27s frame. Using Finite Element Analysis (FEA) in SolidWorks, each design is tested under vertical (25 kg) and horizontal (80 kg) loads, simulating real competition conditions. The Tech United frame shows the best performance, with stress values of 5.8 MPa (vertical) and 180.0 MPa (horizontal), minimal displacements of 0.018 mm and 1.806 mm, and FOS values of 9.5 and 0.3, respectively. It also displays superior stiffness, stability, and a controlled damage zone where deformation is intentionally localized to absorb impact. These characteristics make the Tech United frame the most durable and reliable option, improving overall robot performance and ensuring it can withstand the demands of competition.Frame analysis plays a vital role in the Indonesian Wheeled Soccer Robot Contest (KRSBI-B) by identifying design weaknesses and enabling optimization before building the physical robot. This study compares three frame designs: standard, modified standard, and the Tech United team\u27s frame. Using Finite Element Analysis (FEA) in SolidWorks, each design is tested under vertical (25 kg) and horizontal (80 kg) loads, simulating real competition conditions. The Tech United frame shows the best performance, with stress values of 5.8 MPa (vertical) and 180.0 MPa (horizontal), minimal displacements of 0.018 mm and 1.806 mm, and FOS values of 9.5 and 0.3, respectively. It also displays superior stiffness, stability, and a controlled damage zone where deformation is intentionally localized to absorb impact. These characteristics make the Tech United frame the most durable and reliable option, improving overall robot performance and ensuring it can withstand the demands of competition
Fine-Tuned Transformer Models for Keyword Extraction in Skincare Recommendation Systems
No full textThe skincare industry in Indonesia is experiencing rapid growth, with projected revenues reaching nearly 40 billion rupiah by 2024 and expected to continue to increase. The large number of products in circulation makes it difficult for consumers to find products that suit their needs. In this context, a text-based recommendation system that utilizes advances in Natural Language Processing (NLP) technology is a promising solution. This research aims to develop a skincare product recommendation system based on user needs by applying the DistilBERT model, which is specifically fine-tuned with text in the skincare recommendation domain to perform keyword extraction. The resulting keywords are then used as parameters to provide recommendations by using co-occurrence as well as using a modification of Jaccard Similarity to assess the suitability between the content and benefits of the product and user preferences. The trained extraction model achieved the best performance with a micro F1-score of 0.96 at the token level and an exact match rate of 74.25% at the entity level. The evaluation of the recommendation system showed excellent results, with an nDCG value of 0.96 and a user satisfaction rate (CSAT) of 91.9%