Background: The ratio of ophthalmologists to the population in Indonesia is extremely low, leading to unequal access to eye care and making it difficult for the public to obtain early information regarding eye symptoms.
Aims: This study aims to develop and validate an efficient, domain-specific eye care chatbot using the Llama 3.2 1B Instruct model, focusing on an optimal fine-tuning pipeline for limited hardware constraints.
Methods: Utilizing an Experimental Research (Model Development) design, a QLoRA 4-bit quantization technique was applied on a Google Colab T4 GPU (15.64GB VRAM). The training dataset comprised 16,742 samples from Kaggle Eye Care and MedQuad Indonesian Translation, with a subset of 5,000 samples used for experiments.
Result: After training for 3 epochs (750 steps), the Training Loss decreased from 1.3394 to 0.7188 (46.3% improvement), while Perplexity reached 2.3620, categorized as excellent. The model maintained stability with a final gap of 0.1407 between training and validation loss.
Conclusion: The Meta-Llama 3.2 1B Instruct model, fine-tuned with 4-bit QLoRA, is highly effective for building a domain-specific healthcare chatbot, successfully operating within an 8GB VRAM limit while maintaining high generalization capabilities.

Background: The ratio of ophthalmologists to the population in Indonesia is extremely low, leading to unequal access to eye care and making it difficult for the public to obtain early information regarding eye symptoms.
Aims: This study aims to develop and validate an efficient, domain-specific eye care chatbot using the Llama 3.2 1B Instruct model, focusing on an optimal fine-tuning pipeline for limited hardware constraints.
Methods: Utilizing an Experimental Research (Model Development) design, a QLoRA 4-bit quantization technique was applied on a Google Colab T4 GPU (15.64GB VRAM). The training dataset comprised 16,742 samples from Kaggle Eye Care and MedQuad Indonesian Translation, with a subset of 5,000 samples used for experiments.
Result: After training for 3 epochs (750 steps), the Training Loss decreased from 1.3394 to 0.7188 (46.3% improvement), while Perplexity reached 2.3620, categorized as excellent. The model maintained stability with a final gap of 0.1407 between training and validation loss.
Conclusion: The Meta-Llama 3.2 1B Instruct model, fine-tuned with 4-bit QLoRA, is highly effective for building a domain-specific healthcare chatbot, successfully operating within an 8GB VRAM limit while maintaining high generalization capabilities.

Background: Traffic congestion in Badung Regency, Bali, presents complex challenges for transportation management in a strategic tourism area, necessitating intelligent monitoring methods that surpass conventional surveys.
Aim: This study develops a Hybrid Machine Learning framework that integrates unsupervised and supervised learning techniques to map spatial patterns and precisely predict congestion anomalies.
Methods: Initially, the K-Means Clustering algorithm validated the segmentation of traffic data into five optimal clusters (k=5) based on Elbow and Silhouette Score tests, which subsequently served as the ground truth for data labeling. Furthermore, a comparative evaluation of Random Forest, XGBoost, and Gradient Boosting models was conducted to determine the best predictive performance.
Result: Experimental results identified Random Forest with strict pruning parameters (max_depth=3) as the optimal model, balancing accuracy and generalization stability while effectively avoiding overfitting through learning curve analysis. This model achieved a testing accuracy of 92.00% and a cross-validation score of 89.69%, with the Travel Time Index (TTI) and average speed identified as the primary determinants.
Conclusion: Spatial visualization demonstrated high consistency between the model's prediction maps and actual field conditions, confirming that this approach is effective as a foundation for an Early Warning System to support data-driven traffic management policies in Bali.

Background: The legislative proposal for the revision of the Indonesian Military Draft Bill (RUU TNI) has incited significant public controversy, largely articulated in digital arenas like YouTube. This revision fuels widespread apprehension about the potential revival of the military's dual-function role (Dwifungsi ABRI) and its potential detriment to civilian supremacy.
Aims: This study aims to quantify and analyze public sentiment towards the RUU TNI, as expressed in Indonesian YouTube comments, using a Deep Learning approach combining FastText embedding and Long Short-Term Memory (LSTM) classification.
Methods: A dataset of 520 Indonesian comments was collected and manually labeled into three classes: positive, neutral, and negative. The methodology included comprehensive text preprocessing, feature extraction via FastText word embeddings (300 dimensions), and sentiment classification using the LSTM architecture. The model was rigorously evaluated using a confusion matrix across standard metrics, including accuracy, precision, recall, and F1-score.
Result: The dataset exhibited significant class imbalance, dominated by negative sentiment (42.31%). The optimal LSTM configuration, tested with Stratified K-Fold Cross Validation, yielded an accuracy of 42.39%, a precision of 43%, a recall of 47%, and an F1-score of 60%. Topic modeling via LDA revealed dominant themes criticizing the bill's quality, state corruption, and legislative integrity. The low accuracy, barely surpassing the baseline, suggests that the model struggled with the limited and imbalanced data.
Conclusion: Public sentiment regarding the RUU TNI is strongly negative and critical, reflecting deep concerns about democracy and state institutions. While the FastText-LSTM pipeline was established, its classification performance was severely constrained by the small, highly imbalanced dataset and the complex nature of political discourse. Future research must utilize advanced Transformer models (e.g., IndoBERT) and larger, balanced datasets to achieve meaningful classification performance on this critical socio-political domain.