Jurnal Politeknik Negeri Batam (PoliBatam)
Not a member yet
3001 research outputs found
Sort by
Sentiment Analysis of Economic Policy Comments on YouTube Using Ensemble Machine Learning
No full textPublic sentiment analysis of economic policies is increasingly important in the digital age, as social media platforms have become the main arena for public discussion. This study analyzes YouTube comments related to Tom Lembong\u27s economic policies to address the lack of policy sentiment analysis tools in Indonesian. A dataset containing 1,029 comments was collected and systematically processed using normalization, stop word removal, and stemming techniques tailored to Indonesian. To overcome data scarcity and class imbalance, advanced data augmentation methods—synonym replacement, random insertion, and random deletion—were applied, expanding the dataset to 2,169 samples. Feature extraction used TF-IDF vectorization (unigram, bigram, trigram) and CountVectorizer, followed by an 80:20 split into training and testing sets. Several machine learning algorithms, including Support Vector Machine (SVM), Logistic Regression, Random Forest, Gradient Boosting, and Naïve Bayes, were evaluated with hyperparameter tuning through grid search. The results showed that SVM with TF-IDF bigrams achieved the best performance (accuracy: 96.08%, F1-score: 96.03%). Class-level evaluation showed high performance for negative sentiment (F1-score: 0.97) and positive sentiment (F1-score: 0.97), while neutral sentiment was more challenging (F1-score: 0.90) due to ambiguity, sarcasm, and fewer samples. The ensemble model, which combines several optimized SVM variants with soft voting, achieved robust and stable performance (accuracy and F1-score: 95.16%). These findings confirm the effectiveness of the ensemble-based approach for Indonesian sentiment analysis, while providing valuable insights into public perceptions of economic policy in the digital space.Public sentiment analysis of economic policies is increasingly important in the digital age, as social media platforms have become the main arena for public discussion. This study analyzes YouTube comments related to Tom Lembong\u27s economic policies to address the lack of policy sentiment analysis tools in Indonesian. A dataset containing 1,029 comments was collected and systematically processed using normalization, stop word removal, and stemming techniques tailored to Indonesian. To overcome data scarcity and class imbalance, advanced data augmentation methods—synonym replacement, random insertion, and random deletion—were applied, expanding the dataset to 2,169 samples. Feature extraction used TF-IDF vectorization (unigram, bigram, trigram) and CountVectorizer, followed by an 80:20 split into training and testing sets. Several machine learning algorithms, including Support Vector Machine (SVM), Logistic Regression, Random Forest, Gradient Boosting, and Naïve Bayes, were evaluated with hyperparameter tuning through grid search. The results showed that SVM with TF-IDF bigrams achieved the best performance (accuracy: 96.08%, F1-score: 96.03%). Class-level evaluation showed high performance for negative sentiment (F1-score: 0.97) and positive sentiment (F1-score: 0.97), while neutral sentiment was more challenging (F1-score: 0.90) due to ambiguity, sarcasm, and fewer samples. The ensemble model, which combines several optimized SVM variants with soft voting, achieved robust and stable performance (accuracy and F1-score: 95.16%). These findings confirm the effectiveness of the ensemble-based approach for Indonesian sentiment analysis, while providing valuable insights into public perceptions of economic policy in the digital space
Enhancing Liver Cirrhosis Staging Accuracy using Optuna-Optimized TabNet
No full textLiver cirrhosis is a progressive chronic disease whose early detection poses a clinical challenge, making accurate severity staging crucial for patient management. This research proposes and evaluates a TabNet deep learning model, specifically designed for tabular data, to address this challenge. In the initial evaluation, a baseline TabNet model with its default configuration achieved a baseline accuracy of 65.11% on a public clinical dataset. To enhance performance, hyperparameter optimization using Optuna was implemented, which successfully increased the accuracy significantly to 70.37%, with precision, recall, and F1-score metrics each reaching 70%. The model\u27s discriminative ability was also validated as reliable in multiclass classification through AUC metric evaluation. In addition to accuracy improvements, the model\u27s interpretability was validated through the identification of key predictive features such as Prothrombin and Hepatomegaly, which align with clinical indicators. This study demonstrates that Optuna-optimized TabNet is an effective and interpretable approach, possessing significant potential for integration into clinical decision support systems to support a more precise diagnosis of liver cirrhosis
Identification of Source Code Plagiarism Using a Natural Language Processing (NLP) Approach Based on Code Writing Style Analysis
No full textSource code plagiarism identificatio requires a system capable of identifying semantic similarity rather than mere textual resemblance. This study utilized a dataset of 1,000 source code files, which after cleaning resulted in 996 individual code samples collected from GitHub repositories. The dataset included various programming languages (Python, Java, JavaScript, TypeScript, C++), divided into 697 training data, 149 validation data, and 149 testing data. The model employed was CodeBERT, configured with a hidden size of 768, 12 layers, and 12 attention heads. CodeBERT generated vector embeddings for each code sample, which were then projected by a Siamese Network to calculate cosine similarity between code pairs. Testing used a threshold of 0.80 to classify plagiarism. The identification results achieved an accuracy of 96.4%, precision of 95.2%, recall of 97.8%, F1-score of 96.4%, and an error rate of 4.6%. The system produced similarity scores and status labels of “plagiarism detected” or “not detected,” demonstrating the effectiveness of the CodeBERT-based approach for adaptive and intelligent code similarity identificatio
Comparing Machine Learning Models for Sentiment Analysis of Tokopedia Reviews
No full textThis study presents a comparative evaluation of machine learning models for sentiment analysis on Tokopedia user reviews written in the Indonesian language. The objective is to assess the effectiveness of three algorithms—Support Vector Machine (SVM), Random Forest (RF), and Multilayer Perceptron (MLP)—in classifying customer sentiments extracted from Tokopedia reviews on Google Play Store. The dataset, collected between January and October 2025, consists of 10,236 unique entries after preprocessing, which included text cleaning, case folding, tokenization, stopword removal, normalization using a verified Indonesian word normalization dictionary, and optional stemming with the Sastrawi library. The reviews were divided into positive and negative categories based on rating polarity (4–5 stars as positive; 1–2 stars as negative).Each model was evaluated using both hold-out validation (80:20 split) and 5-fold cross-validation, employing metrics such as accuracy, precision, recall, and F1-score. Experimental results indicate that the SVM achieved the highest accuracy of 0.88, outperforming Random Forest (0.85) and MLP (0.83). These findings demonstrate that SVM performs more robustly on sparse TF-IDF vector features and is more resistant to noise within informal Indonesian expressions. The research further discusses the linguistic challenges inherent in Indonesian sentiment analysis, including code-mixing, abbreviations, and non-standard words, while proposing preprocessing strategies to mitigate them.The outcomes of this study contribute to enhancing the reliability of sentiment-based decision support systems in Indonesian e-commerce platforms. The methodological framework developed here can serve as a baseline for future work involving hybrid or deep-learning approaches such as LSTM or IndoBERT for improved contextual understanding.This study presents a comparative evaluation of machine learning models for sentiment analysis on Tokopedia user reviews written in the Indonesian language. The objective is to assess the effectiveness of three algorithms—Support Vector Machine (SVM), Random Forest (RF), and Multilayer Perceptron (MLP)—in classifying customer sentiments extracted from Tokopedia reviews on Google Play Store. The dataset, collected between January and October 2025, consists of 10,236 unique entries after preprocessing, which included text cleaning, case folding, tokenization, stopword removal, normalization using a verified Indonesian word normalization dictionary, and optional stemming with the Sastrawi library. The reviews were divided into positive and negative categories based on rating polarity (4–5 stars as positive; 1–2 stars as negative).Each model was evaluated using both hold-out validation (80:20 split) and 5-fold cross-validation, employing metrics such as accuracy, precision, recall, and F1-score. Experimental results indicate that the SVM achieved the highest accuracy of 0.88, outperforming Random Forest (0.85) and MLP (0.83). These findings demonstrate that SVM performs more robustly on sparse TF-IDF vector features and is more resistant to noise within informal Indonesian expressions. The research further discusses the linguistic challenges inherent in Indonesian sentiment analysis, including code-mixing, abbreviations, and non-standard words, while proposing preprocessing strategies to mitigate them.The outcomes of this study contribute to enhancing the reliability of sentiment-based decision support systems in Indonesian e-commerce platforms. The methodological framework developed here can serve as a baseline for future work involving hybrid or deep-learning approaches such as LSTM or IndoBERT for improved contextual understanding
User Requirement Recommendation Model for Waste Reporting Platforms Based on UX Topics and Sentiment Analysis
No full textWaste management remains a critical issue in Indonesia, as emphasized in the RPJMN 2025–2029. Ineffective collection and processing services, coupled with limited public participation, continue to hinder progress. Meanwhile, social media has emerged as a primary channel for citizens to express complaints and reports on waste, yet the unstructured nature of comments poses challenges for integration into official reporting systems. This study proposes a user requirements recommendation model based on social media data by integrating sentiment analysis, topic modeling, and rule-based recommendation. Data were collected from YouTube and TikTok comments. Sentiment classification was performed using Support Vector Machine (SVM), while Latent Dirichlet Allocation (LDA) was employed for topic modeling, with results mapped onto the UX Honeycomb dimensions. Recommendation rules were then formulated by combining sentiment polarity with UX dimensions. The SVM model achieved an average accuracy of 87.5% with balanced precision, recall, and F1-score. LDA produced 15 coherent topics, which were distributed across seven UX dimensions. The integration revealed that the main user requirements include transparency in report follow-up through real-time notifications and clear status updates. Additional recommendations involve simplifying the reporting process, providing auto-fill features, improving visual design, and establishing a user appreciation system. The findings demonstrate the potential of leveraging social media comments to systematically capture user requirements, offering practical insights for developers to design waste reporting platforms that are effective, user-friendly, and responsive to community expectations
Comparison of Sarima and Exponential Smoothing Methods in Forecasting Exchange Rates for Farmers in Central Java Province
No full textThis study compares the performance of the SARIMA and Exponential Smoothing (Holt-Winters) models in forecasting the Farmer Exchange Rate (NTP) for Central Java Province from 2016 to 2025. The monthly statistical data used was obtained from the Central Java Provincial Statistics Agency. The models were evaluated using MAE (Mean Absolute Error) and RMSE (Root Mean Squared Error) on test data for the period January 2016 to September 2025, while forecasting was carried out from October 2025 to December 2027. The results show that the SARIMA (1,1,1) (1,1,1,12) model has an MAE of 6.94 and an RMSE of 7.88, indicating that the model can make accurate predictions with few errors. However, the Exponential Smoothing model has a lower MAE and RMSE, implying that this model is more accurate in forecasting long-term NTP. Both models show comparable seasonal trends, with Exponential Smoothing being more stable and sensitive to seasonal changes. This study also proposes the use of alternative forecasting approaches, such as ARIMAX, VAR, or machine learning to improve the accuracy of future forecasts. The results of this study can be used to develop agricultural policies that maintain food price stability, improve farmer welfare, and predict future inflation fluctuations
Uji Akurasi Peta Foto Udara Skala 1:5000 Sebagai Peta Dasar Berdasarkan Juknis PTSL Tahun 2023 (Studi Kasus : Tanjung Piayu, Batam, Kepualauan Riau)
No full textPerkembangan teknologi saat ini sangatlah pesat di bidang pemetaan. Salah satunya pemetaan dengan metode fotogrametri menggunakan wahana Unmanned Aerial Vehicle (UAV) atau disebut Pesawat Udara Nir Awak (PUNA). Petunjuk Teknis Pendaftaran Tanah Sistematis Lengkap (PTSL) Kementrian Agraria dan Tata Ruang / Badan Pertanahan Nasional (ATR/BPN) Nomor 3/Juknis-HK.02/III/2023 Bab II/F./1./e./5) Tabel 2 Ketentuan Spesifikasi Teknis Hasil Pekerjaan Peta Foto menerangkan bahwa aspek resolusi (GSD) harus dengan nilai ≤ 0,15 meter dan Ketelitian Horizontal (CE90) harus dengan nilai ≤ 0,5 meter. Pada penelitian ini menerapkan pengukuran dan pemetaan menggunakan metode fotogrametri dengan teknik pemotretan tegak lurus. Tujuan penelitian ini yaitu untuk mengetahui perbandingan uji akurasi berdasarkan standar ketelitian Juknis PTSL Tahun 2023 BPN (Badan Pertanahan Nasional) pada ortofoto. Hal tersebut menjadi acuan / standar ketelitian yang dipakai dalam pengujian akurasi data terhadap pemotretan foto udara dengan skala 1:5000. Area pemotretan yang digunakan untuk uji akurasi sekitar ± 70 ha, dengan persebaran titik GCP berjumlah 10 titik yang digunakan untuk koreksi geometrik. Pengujian akurasi data dengan cara menganalisis ketelititan nilai RMSE (Root Mean Square Error) dan menganalisis hasil perbandingan pengukuran terrestris dengan pengamatan GPS (Global Positioning System) metode Statik. Hasil uji akurasi data pada ortofoto berdasarkan standar ketelitian horizontal (CE90) yang dikeluarkan BPN untuk peta foto memenuhi syarat. Berdasarkan standar resolusi yang dikeluarkan BPN untuk peta foto memenuhi syarat. Maka dapat disimpulkan bahwa hasil resolusi dan perbandingan ketelitian horizontal (CE90) dapat dijadikan peta dasar BPN untuk pekerjaan PTSL
Performance Comparison of Fuzzy Mamdani and Sugeno in Curtain and Lighting Control
No full textEfficient room lighting is essential for user comfort and health. This study implements and compares Fuzzy Mamdani and Sugeno methods in an automatic curtain and lighting control system based on Arduino Uno. The system responds to light intensity changes by adjusting curtains and a dimmable lamp according to LDR sensor values. Tests were conducted under various conditions to evaluate performance. Results show that Sugeno has a faster average response time of 13,898 seconds compared to Mamdani\u27s 15,158 seconds, with a difference of 1,26 seconds. Sugeno also demonstrates better accuracy: for curtain control, it achieved MSE of 347,9368872 and RMSE of 18,65306643, while Mamdani recorded MSE of 570,71862 and RMSE of 23,889718. For dimmer control, Sugeno obtained MSE of 1683,99071 and RMSE of 41,0364559, outperforming Mamdani\u27s MSE of 2679,24 and RMSE of 51,7614. This implementation provides an optimal solution for automatic lighting management to enhance energy efficiency and user comfort
Dynamic Modeling and PID Control of a 6-DOF Robotic Arm Using ROS and Gazebo
No full textThis paper presents the dynamic modeling and control evaluation of a six degrees-of-freedom (6-DOF) robotic manipulator. The manipulator was developed in the Robot Operating System (ROS) and Gazebo using a detailed URDF model with complete geometric and inertia parameters. Proportional–Integral–Derivative (PID) controllers were tuned through ROS dynamic reconfiguration and tested under four payloads: 0; 0,19; 0,39; and 0,50 kg. Controller performance was assessed using rise time, settling time, overshoot, and steady-state error. The results show stable responses across all conditions, with no overshoot and near-zero steady-state errors. Increasing payloads generally led to longer rise and settling times, while joints aligned with gravity exhibited faster responses under heavier loads. These findings confirm that properly tuned PID controllers can maintain robust and accurate manipulator performance and demonstrate the effectiveness of ROS–Gazebo as an open-source platform for robotic control experimentation and future integration of adaptive or AI-based methods
Educational Animation Media on the Effects of Ultra-Processed Foods to Children\u27s Cognitive Function: A Systematic Literature Review
No full textThe consumption of ultra-processed food (UPF) poses a serious risk of cognitive impairment to children during their critical brain development phase. Effective educational media is crucial, as animation is a highly engaging and proven medium for children, with 90% of them choosing animation for education in previous research. This study, therefore, aims to systematically review the topic by addressing three key research questions (RQs): (1) To what extent does UPF consumption affect children\u27s cognitive function and health? (2) Why are animations the most effective media for transferring nutritional knowledge? (3) How effective is animation-based educational media in raising children\u27s awareness regarding UPF risks? This study adopts a Systematic Literature Review (SLR) methodology, adhering to the PRISMA guidelines. The systematic literature search gathered scientific articles from reputable world libraries. The results confirm that UPF consumption is a significant factor contributing to cognitive decline and obesity due to its high sugar and low micronutrient content. Conversely, animation is proven effective as it significantly increases children\u27s focus and comprehension. The core conclusion is that educational animation media serves as a strategic and highly effective tool for fostering children\u27s awareness of the dangers of consuming UPF and promoting preventative behavior.Konsumsi makanan ultra-proses (MUP) menimbulkan risiko serius gangguan kognitif pada anak-anak selama fase kritis perkembangan otak mereka. Media edukasi yang efektif sangat penting, karena animasi adalah media yang sangat menarik dan terbukti efektif untuk anak-anak, dengan 90% dari mereka memilih animasi untuk pendidikan dalam penelitian sebelumnya. Oleh karena itu, penelitian ini bertujuan untuk meninjau topik ini secara sistematis dengan menjawab tiga pertanyaan penelitian (PP) utama: (1) Sejauh mana konsumsi MUP memengaruhi fungsi kognitif dan kesehatan anak? (2) Mengapa animasi merupakan media yang paling efektif untuk mentransfer pengetahuan gizi? (3) Seberapa efektif media edukasi berbasis animasi dalam meningkatkan kesadaran anak-anak tentang risiko MUP? Penelitian ini mengadopsi metodologi Tinjauan Literatur Sistematis (SLR), yang mematuhi pedoman PRISMA. Pencarian literatur sistematis mengumpulkan artikel ilmiah dari perpustakaan dunia terkemuka. Hasilnya menegaskan bahwa konsumsi MUP adalah faktor signifikan yang berkontribusi terhadap penurunan kognitif dan obesitas karena kandungan gula yang tinggi dan mikronutrien yang rendah. Sebaliknya, animasi terbukti efektif karena secara signifikan meningkatkan fokus dan pemahaman anak-anak. Kesimpulan utamanya adalah bahwa media animasi edukasi berfungsi sebagai alat yang strategis dan sangat efektif untuk menumbuhkan kesadaran anak-anak tentang bahaya mengonsumsi MUP dan mendorong perilaku pencegahan