Wayan Oger Vihikan, Wayan Oger
Udayana University

Published : 8 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 8 Documents
Search

Cryptocurrency Price Predictor: Aplikasi Prediksi Harga Crypto Dengan Perbandingan Metode ARIMA, LSTM Dan SARIMAX Salsabila, Archels Ramadhany; Suarjaya, I Made Agus Dwi; Vihikan, Wayan Oger
JITTER : Jurnal Ilmiah Teknologi dan Komputer Vol 5 No 2 (2024): JITTER, Vol.5, No.2, August 2024
Publisher : Program Studi Teknologi Informasi, Fakultas Teknik, Universitas Udayana

Show Abstract | Download Original | Original Source | Check in Google Scholar

Abstract

Cryptocurrency telah menjadi inovasi teknologi informasi di bidang keuangan yang paling menarik perhatian investor, peneliti dan masyarakat lain dalam beberapa tahun terakhir di seluruh dunia. Penelitian ini bertujuan untuk mengembangkan aplikasi prediksi harga cryptocurrency dengan hasil perbandingan kinerja tiga metode prediksi yang populer: Autoregressive Integrated Moving Average (ARIMA), Long Short-Term Memory (LSTM), dan Seasonal Autoregressive Integrated Moving Average with Exogenous Factors (SARIMAX). Dalam penelitian ini, dikumpulkan data historis harga cryptocurrency dari sumber terpercaya kemudian melatih dan membangun metode-metode tersebut untuk memprediksi harga cryptocurrency di 2 tahun ke depan. Kinerja masing-masing metode dievaluasi berdasarkan metrik akurasi prediksi Mean Absolute Percentage Error (MAPE) dan Root Mean Squared Error (RMSE) pada LSTM. Hasil penelitian ini menunjukkan metode LSTM lebih unggul dalam menangkap dependensi jangka panjang dan pola non-linear yaitu mendapatkan rata-rata akurasi 5,1%.
Sentiment Analysis of X (Twitter) Comments on The Influence of South Korean Culture in Indonesia Savitri, Putu Rheya Ananda; Suarjaya, I Made Agus Dwi; Vihikan, Wayan Oger
Journal of Information System and Informatics Vol 6 No 2 (2024): June
Publisher : Universitas Bina Darma

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.51519/journalisi.v6i2.749

Abstract

Hallyu or Korean wave refers to the phenomenon of South Korean values and culture spreading to other countries, ultimately influencing global culture. South Korean culture, such as K-pop music, dramas, films, fashion, food, and lifestyle, has gained popularity in Indonesia since 2002. Because South Korean culture influences many aspects of life in Indonesia, responses to this Korean wave are widely discussed in social media, especially through X (Twitter) ranging from positive sentiment to negative sentiment. To gain a more in-depth and detailed understanding of public opinion, a classification process was conducted on the social media platform X (Twitter) using a deep learning algorithm based on the CNN method. The results of this classification provide more accurate and informative insight into the attitudes, opinions, and reactions of the Indonesian people towards the influence of South Korean culture in this country. The research was conducted using 717,998 tweet data resulting in an accuracy of 79%.
Sentiment Analysis of Indonesian Citizens on Electric Vehicle Using FastText and BERT Method Wijaya, Darryl Rayhan; Sasmitha, Gusti Made Arya; Vihikan, Wayan Oger
Journal of Information System and Informatics Vol 6 No 3 (2024): September
Publisher : Universitas Bina Darma

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.51519/journalisi.v6i3.784

Abstract

Electric vehicles have become one of the most important innovations in the automotive industry in recent years. This is not only related to technological developments, but also to its significant impact on the environment and lifestyle of global society. Lot of people do not know about the benefit of using electric vehicles for our environment. The transition from conventional vehicles to electric vehicles can really make our environment healthier and also reducing the pollution. At the same time, debates and feelings about electric vehicles continue to grow around the world. This study aims to understand the dynamics of people's feelings and opinions about electric vehicles through sentiment analysis using the FastText and IndoBERT methods. FastText is an efficient text classification and representation learning method developed by Facebook's AI Research (FAIR) lab. IndoBERT is a pre-trained language model specifically designed for the Indonesian language, leveraging the Bidirectional Encoder Representations from Transformers (BERT) architecture. By analyzing a total of 119,310 data from January 2020 to June 2023, the tweets data were categorized into negative, neutral, and positive classes. Model yielded the highest accuracy of 82.5% using IndoBERT method. The results outcomes positive perceptions of electric vehicles among Indonesian citizen with a percentage of 58%. By carrying out this research, it is hoped that it can produce quality information for producers, the community and the government in developing and advancing public interest in purchasing electric vehicles considering the very positive impact they have on the surrounding environment.
Indonesian Health Question Multi-Class Classification Based on Deep Learning Vihikan, Wayan Oger; Trisna, I Nyoman Prayana
Journal of Information System and Informatics Vol 6 No 3 (2024): September
Publisher : Universitas Bina Darma

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.51519/journalisi.v6i3.838

Abstract

The health online forum is commonly used by Indonesian to ask questions related to diseases. A well-known example, Alodokter, has hundreds of thousands of health questions which are assigned to certain topics. Building a model to classify questions into a topic is important for better organization and faster response by relevant health professionals. This research experimented on 20 deep learning methods from RNN, CNN, and IndoBERT with different configurations to see the performance of each model when classifying questions into six different most common diseases that cause death in Indonesia. The results show the majority of the model can outperform the SVM as baseline. Bidirectional RNN such BiLSTM and BiGRU combined with CNN show a good metric score even though a certain version of the IndoBERT model generally outperforms all the other models.
Baby Cry Classification Using Ensemble Learning and Whisper Method Comparison Dharmawan, I Putu Yogi Prasetya; Suarjaya, I Made Agus Dwi; Vihikan, Wayan Oger
Journal of Applied Informatics and Computing Vol. 9 No. 2 (2025): April 2025
Publisher : Politeknik Negeri Batam

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30871/jaic.v9i2.9167

Abstract

Baby cry classification is an important topic in Machine Learning, especially in the healthcare field, as crying is the primary form of communication for infants to convey their needs or conditions. Many inexperienced parents tend to interpret baby cries in a limited way, even though each cry has unique characteristics that represent specific needs such as hunger, discomfort, sleepiness, flatulence, and abdominal pain. With the advancement of technology, identification of baby cries can now be done automatically through AI-based applications, but the implementation is still limited. This study compares the performance of ensemble learning methods, namely Random Forest and XGBoost, with the Whisper model in classifying baby cries. The results show that the Whisper-small model has the best performance with precision 0.9115 and recall 0.9007, followed by XGBoost with slightly degraded performance after hyperparameter optimization. Random Forest showed the lowest performance among the three models. Transformer-based models such as Whisper-small proved to be superior in capturing the complex patterns of infant cries, compared to tree-based models. These findings indicate the great potential of accurate and reliable models to help parents understand the needs of infants more effectively, thereby improving the quality of infant care.
Implementation of a Telegram-Based Child Consultation Chatbot Using IndoBERT Whurapsari, Gusti Ayu Wahyu; Suarjaya, I Made Agus Dwi; Vihikan, Wayan Oger
Journal of Information System and Informatics Vol 7 No 2 (2025): June
Publisher : Universitas Bina Darma

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.51519/journalisi.v7i2.1079

Abstract

Children’s health and development are crucial aspects that require proper attention from parents. However, many parents lack easy access to immediate consultation regarding their child's health and well-being. To address this issue, this study develops a child consultation chatbot on Telegram using the IndoBERT model. The chatbot utilizes data from Halodoc and Alodokter, structured into an intent-based format with 227 tags, 5,428 patterns, and 278 responses. The dataset undergoes preprocessing, including lowercasing, text cleaning, normalization, stopword removal, and stemming. Four preprocessing scenarios are tested, including the use of term frequency-based stopwords without applying stemming, the use of NLTK stopwords without stemming, the use of term frequency-based stopwords combined with stemming, and the use of NLTK stopwords combined with stemming. The best model, trained with an 80:20 training-validation split using term frequency-based stopwords without stemming, achieves 98% accuracy, 98.5% F1-score, 98.9% precision, and 98.5% recall. The chatbot successfully classifies user intent and ensures structured interactions through a confidence-based response mechanism. This research demonstrates that an IndoBERT-based chatbot can effectively assist parents in obtaining quick and relevant information regarding their children's health and development.
Web-Based Makeup Recommendation System Using Hybrid Filtering Utami, Putu Mia Setya; Trisna, I Nyoman Prayana; Vihikan, Wayan Oger
Journal of Applied Informatics and Computing Vol. 9 No. 3 (2025): June 2025
Publisher : Politeknik Negeri Batam

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.30871/jaic.v9i3.9339

Abstract

The increasing use of makeup products in the modern era, driven by evolving beauty trends and e-commerce accessibility, presents challenges in selecting products suited to individual skin types and conditions. A recommendation system addresses this issue by enhancing selection efficiency. This study explores the implementation of Content-Based Filtering (CBF) using TF-IDF and Cosine Similarity, Collaborative Filtering (CF) with Singular Value Decomposition (SVD), and a Hybrid Filtering approach integrating both methods through Weighted Hybrid techniques. The system's performance is evaluated across two user scenarios: new users (without prior ratings) and old users (with rating history). The evaluation method includes Precision, Normalized Discounted Cumulative Gain (NDCG), and accumulation of the best scenario based on user opinion. Results show that Hybrid Filtering outperforms CBF and CF, with notable differences between user groups. For new users, 32% prefer Scenario 1, which emphasizes CBF, achieving 80.8% Precision and 89.73% NDCG. For old users, 23% favor Scenario 2, attaining 83.4% Precision and 90.31% NDCG.
Classifying Indonesian Hoax News Titles with SVM, XGBoost, and BiLSTM Trisna, I Nyoman Prayana; Putra, I Made Wiraharja Jaya; Vihikan, Wayan Oger
IJCCS (Indonesian Journal of Computing and Cybernetics Systems) Vol 19, No 4 (2025): October
Publisher : IndoCEISS in colaboration with Universitas Gadjah Mada, Indonesia.

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.22146/ijccs.106608

Abstract

This study investigates the automated detection of hoaxes related to President Jokowi in Indonesian news by analyzing only news titles, aiming for efficient detection and reduced traffic to harmful websites. We compared the performance of traditional (SVM, XGBoost) and deep learning (BiLSTM) algorithms, with and without Synthetic Minority Over-sampling Technique (SMOTE) to address class imbalance in a dataset scraped from trusted news sources (CNN Indonesia, Detik News) and a fact-checking platform (turnbackhoax.id). The results indicate that BiLSTM generally outperformed SVM and XGBoost, demonstrating the potential of deep learning for this task. However, applying SMOTE negatively impacted BiLSTM's performance, suggesting overfitting. Notably, precision consistently exceeded recall across all models, indicating high reliability in identifying hoaxes but a potential for missing a significant number of actual hoaxes. This highlights a trade-off between avoiding false positives and ensuring comprehensive detection. The findings also suggest that language-specific characteristics influence algorithm effectiveness. This research contributes to developing efficient and accurate tools for combating misinformation in the Indonesian online environment, emphasizing the importance of title-based analysis and careful consideration on data balancing.